Telcos are eager for new revenue streams, and AI is today’s hot market, so it’s not a surprise to see the telco push. But most of what the telcos are selling here is someone else’s AI product. Each telco is building a switchboard, in essence, which will provide one contract, one bill and a routing layer that sends each prompt to the model that fits best. The telcos’ pitch is the wrapper, and this is where they hope to reap good margins.

Subscribe now

Singtel resells Anthropic and OpenAI under one contract

RE:AI, the sovereign AI cloud business inside Singtel’s Digital InfraCo, calls its product AI Token-as-a-Service (TaaS). The launch release says that open-weight models such as Qwen, Mistral, GLM, Kimi and MiniMax are running on RE:AI’s own infrastructure. The product page adds that closed-weight models “from Anthropic, OpenAI, Bedrock and Azure” are resold under one contract and billed per token, “with our margin applied.”

The telco’s main value add is the routing functionality. TaaS “automatically matches workloads to the most suitable AI models to optimize cost and performance,” according to the release. Sensitive work stays on Singtel’s sovereign infrastructure in Singapore, and less sensitive work can go to non-sovereign clouds. RE:AI already rents out graphics processing units (GPUs) through a GPU-as-a-Service offer. Adding on a token business moves Singtel one layer up, from renting hardware to brokering models. RE:AI promises token-level visibility into usage. Buyers should check that it also shows why each request went to the model it did, because the telco earns different margins on different models.

Indosat puts an in-country model on the same switchboard

Indosat’s Intelligence Suite offers more than 300 open and closed models through one interface. Its catalog lists Claude Sonnet 5, Gemini 2.5 Pro, DeepSeek-V3.2 and Llama 4 Maverick beside Sahabat-AI, a model it hosts in-country. Customers buy tokens as credits and spread them across models, applications and teams.

US-based vendor F5 provides the traffic management function, including load balancing that steers requests across GPUs. Indosat’s top tier is a dedicated private instance with “in-country processing and reserved capacity.” Indosat is targeting 50 businesses at launch, according to the F5 release.

SK Telecom plans to build a token gateway

SK Telecom is a step behind Singtel and Indosat: it has a plan but no product. Chief executive Jung Jae-hun told staff at an 8 October town hall that the company will build a Token Gateway. SK Telecom describes it as a control center that tracks how companies and customers use AI and connects them to the best-suited model and infrastructure.

SK Telecom’s competitive edge in this setup is billing, not AI models, and most telcos can claim the same edge. The gateway will build on about 23 million subscribers and SK Telecom’s existing payment system. China Mobile already lets Shanghai customers pay for tokens through their phone bills, China Daily reported in May. Singtel and Indosat already bill the enterprises they are targeting. SK Telecom’s plan stands out for aiming at consumers as well as companies. Integrated billing from a known entity lowers the friction of a first purchase, even if it doesn’t raise the cost of leaving.

China’s telcos show what a token switchboard costs

China Telecom launched trial token packages on 17 May, and China Mobile and China Unicom followed. China Telecom’s consumer plans amount to $0.09 to $0.15 per million tokens and its business plans at $0.30 to $0.40. China Mobile charges about $0.37 per million tokens in Beijing and Shanghai, according to the same report. China Mobile’s platform connects more than 300 mainstream AI models, including DeepSeek, Qwen, Doubao and GLM.

Comparing China Mobile’s prices to the list prices of the AI model companies is illuminating. Anthropic lists Claude Sonnet 5 at $2 per million input tokens and $10 per million output tokens, the same model Indosat offers. China Mobile’s $0.37 is less than a fifth of Anthropic’s input price and under a 25th of its output price. Its packages don’t split input from output, and a token isn’t a fixed unit: each model family splits text its own way. The gap mostly reflects the models. Qwen and DeepSeek are open-weight models, while Claude Sonnet 5 is a frontier model. China’s packages also hint at how tokens end up when telcos sell them as a utility: in monthly buckets, priced like data.

T-Systems, Deutsche Telekom’s IT services arm, publishes a price for each model it resells, which allows a like-for-like check. For Claude Sonnet 5 it charges $3.31 per million input tokens and $16.56 per million output tokens, about two-thirds above Anthropic’s own list price. Its rate for OpenAI’s GPT-6.1 Sol, which it resells from Microsoft Azure, is 19% above OpenAI’s. For Zhipu’s GLM-5.2, which T-Systems runs in Germany, its input price is 19% above Zhipu’s own and its output price 11% below. T-Systems’ premium is widest on Claude.

In Singapore, telcos host the open models locally, not the frontier ones

Singtel and Indosat don’t publish per-model rates, but they do describe their offerings. Indosat’s models page sorts its catalog into tiers. Open-weight models such as GLM, Qwen and DeepSeek run on Indosat’s GPUs. Indosat says their data stays “in the nominated jurisdiction.” It describes Anthropic, Google and xAI models as “frontier models reached via the provider.”

A Singtel customer can keep frontier-model data local without a telco, but not through every model maker’s own application programming interface (API). Anthropic’s API lets customers pin processing to the US, at a 10% premium, but not to Singapore. Amazon Bedrock began running Claude Sonnet 5 inside its Singapore region on 29 September, and Anthropic’s pricing page says such regional endpoints cost 10% more than global ones. OpenAI adds 10% for data residency, which it offers in Singapore to approved enterprise customers. Indosat reaches frontier models through the provider.

RE:AI’s product page draws the same line for Singtel. Its own infrastructure runs the open-weight models, so prompts to them stay in Singapore. For Claude and GPT, the page says only that they are “resold under one contract,” which appears to be a pass-through to the provider. A prompt to Claude would then run on the model maker’s or its cloud provider’s servers, not Singtel’s. Singtel hasn’t said whether its Singapore offer covers them.

On open-weight models the sovereignty claim holds, but running a model in-country costs more even when a buyer goes straight to its developer. Alibaba sells Qwen from its own Singapore deployment, at $1.60 per million input tokens for Qwen Max. That is 4.6 times its Beijing price of $0.35. Z.ai lists GLM-5.2 at $1.40 per million input tokens. Its pricing page doesn’t say where the model runs. A telco that hosts GLM or Qwen in Singapore or Jakarta has to price against these developer rates.

That makes the wrapper two businesses. On closed models such as Claude, the telco earns a spread on the model maker’s list price. With open models, the telco runs the model on its own GPUs, so its margin depends on how busy it keeps them. Singtel and Indosat are running both at once.

A local telco can make a Chinese model easier to buy

Singtel’s launch release gives five examples of open-weight models, and four come from Chinese developers: Qwen, GLM, Kimi and MiniMax. RE:AI runs them on its own infrastructure in Singapore. T-Systems runs Zhipu’s GLM and Alibaba’s Qwen models on its T Cloud in Germany, its price list shows.

That changes the security question for buyers wary of Chinese AI. An open-weight model running on a local telco’s servers sends prompts and data to that telco, not to the developer’s own cloud service in China. The customer gets a cheap model and a supplier it already knows and can hold to local rules. I would personally trust my local telco more than Anthropic or OpenAI.

Local hosting doesn’t settle every concern. It controls where the data goes, but not how the model was trained or what it will and won’t answer.

The wrapper has to be worth paying for

The obvious objection is that an enterprise can buy Claude or Gemini directly, so why pay a telco in the middle? Singtel, Indosat and SK Telecom aren’t building frontier models themselves, so what exactly are they offering?

What the telcos can sell is everything around the model. That means keeping data in the country, one contract across many model makers, routing that picks a cheaper model when it will do, and billing simplicity. These are all attractive features. Telcos will soon find out if they support prices high enough to deliver a reasonable margin on this new line of business.

Subscribe now

Questions about this post? Contact the author: Matt Walker, Chief Analyst of MTN Consulting (matt@mtn-c.com)

Interested in exploring (some of) the data behind MTN Consulting’s telco AI coverage? Check out our Telco Workforce Monitor, which tracks headcount, labor costs, profitability and AI adoption at 145 telcos. We also track overseas deployments of Chinese AI models by telcos, here: https://genaitelco.com/china-ai-tracker.html.

AT&T isn’t cutting because business is bad. It is cutting while business is on the upswing. While AI is widely viewed as a job killer, that’s not necessarily true in telecom. AT&T now cites its shift from copper to fiber as the main enabler.

Subscribe now

AT&T credited AI with its savings in February, but cited retirement in October

CEO John Stankey set out the savings plan in his February 2026 shareholder letter. AT&T, he said, planned “an additional $4 billion in annual cost savings by the end of 2028 as we leverage AI.” Digital customer transactions and operating leverage came second and third.

On the 21 October call, CFO Desroches repeated the $4B target, but pinned it to the old network instead. “Copper retirement remains an integral part of our overall cost transformation,” he said. Neither executive linked AI to costs, headcount or the separation charge. Stankey’s references to AI were focused on demand. The fiber network, he said, “has us well positioned to be a leader in AI-grade connectivity.” The results release mentions AI once, in his line that “our asset base positions us well for the AI era.”

AT&T does use AI as a tool to reduce its own costs, but has not quantified the savings. It has, however, talked about how it keeps its AI costs down. Chief Data and AI Officer Andy Markus wrote in July that AT&T runs an average of 45 billion tokens a day, which can be a significant expense. Its AI Gateway, he said, is “reducing AI costs as much as 90%, already saving millions.” Markus explained that it does this through an “intelligent, cache-aware routing and governance system that acts as a control layer for all artificial intelligence requests across the company.”

The workforce has halved while AT&T moved from copper to fiber

AT&T employed 130,870 people in 2Q26, 49% fewer than the 256,420 it had at the end of 2011, according to MTN Consulting’s Telco Workforce Monitor. It cut 7,960 jobs in 2025 and another 2,160 in 1H26. It has fiber passing 32 million locations and aims to pass more than 60 million by the end of 2030, according to its 2025 Sustainability Update.

Copper has come up on every AT&T call this year. In January, Stankey said AT&T had stopped selling legacy copper services in 85% of its wire centers, and Desroches said AT&T aims to discontinue a large majority of copper-based services by the end of 2029. In July, Stankey described an orderly turndown of legacy copper services by the end of the decade. Like most telcos, AT&T doesn’t tie any of this to a job count. BT is a similar case, as we showed in a recent post. Deploying more fiber deep in the access network reduced BT’s network faults, and that has supported a smaller Openreach workforce. Like AT&T, though, BT has yet to put a number on what AI saved.

AT&T doesn’t report personnel expenses, and neither do most US telcos. MTN Consulting estimates AT&T’s labor cost from the ratio of its CEO’s pay to its median employee’s pay, plus US Bureau of Labor Statistics (BLS) pay data for the telecom sector. On that estimate AT&T spent $4.8 billion (B) on labor in 2Q26. That works out to $143,300 per employee over the four quarters to 2Q26, 1.8 times the $79,300 average across 40 telcos in the Americas. Revenue per employee is higher too, though, at $972,300 against $619,800.

Source: MTN Consulting’s Telco Workforce Monitor, Figure 13, AT&T versus the Americas average (40 telcos), 2Q21 to 2Q26, annualized, US dollars. AT&T labor cost is an MTN Consulting estimate.

AT&T gives no job count for the $800M

AT&T hasn’t said how many people the charge covers. Its 10-Q describes separation charges only as costs “associated with voluntary and/or strategic offers,” and neither the release nor the call gives a number of employees impacted. For scale, $800M is about a sixth of our estimate of what AT&T spent on labor in 2Q26.

Source: MTN Consulting’s Telco Workforce Monitor, AT&T, Table 02, real quarterly figures, US$ millions except employees. Labor cost is an MTN Consulting estimate.

The fiber switch also cuts AT&T’s power bill

Fiber needs less electricity than copper, and AT&T addresses this in its 2025 Sustainability Update. It reports saving 660,000 megawatt-hours (MWh) since 2024 by moving its network from copper to fiber. For the five-year period from 2024 through 2028, AT&T expects the fiber transition to reduce energy use by 1.06 million MWh and its emissions by about 740,000 metric tons of carbon dioxide equivalent (CO2e). AT&T’s total energy use, by its reported KPIs, was 15.0 million MWh in 2025, up from 14.8 million MWh in 2024. The fiber shift may not necessarily result in lower energy consumption, though: AT&T says the AI era “will also drive increased power usage.”

AT&T will also spend $1B a year less on its network

AT&T also talked about capex projections on its recent call. The company now expects capital investment of $22B to $23B a year in 2027 and 2028, down from the $23B to $24B it guided in July. Desroches said $300M to $400M of the reduction comes from selling AT&T Mexico’s wireless business. “The balance is from the efficiencies I described,” he said.

AT&T’s quarterly headcount, revenue, capex, labor cost and profit since 1Q11 sit in the Telco Workforce Monitor, alongside the same series for MTN Consulting’s 145-telco universe. Headcount, revenue and capex are free to explore with registration. Full labor-cost and profit history for the 72-telco panel, including the AT&T charts above, is $799 as a one-time purchase of one edition.

Subscribe now

Questions about this post? Contact the author: Matt Walker, Chief Analyst of MTN Consulting (matt@mtn-c.com)

Interested in exploring (some of) the data behind MTN Consulting’s telco AI coverage? Check out our Telco Workforce Monitor, which tracks headcount, labor costs, profitability and AI adoption at 145 telcos:

Open the Telco Workforce Monitor

That’s the job Telkom Indonesia and IOH do every day, and both run AI in live operations. Together they account for 77% of Indonesian telco revenue in 2Q26, and XLSMART, the only other significant player, accounts for 21%, according to MTN Consulting’s Global Telco Market Tracker. MTN Consulting has profiled both for its Telecom AI & Automation (TAIA) research. TAIA scores deployment depth on six criteria, each from 0 to 3, and the profiles of 7 and 9 October 2026 give IOH and Telkom 12 of 18 each.

Indonesia rarely comes up when the industry lists its AI leaders. It should. Both companies stay well above the global telco average on profitability, run leaner than most peers, and have turned TM Forum’s awards stage into a regular stop. The identical scores hide two different stated strategies. IOH calls its strategy AI North Star, with three pillars: an AI-native telco, an AI cloud business, and a nation builder for Indonesia’s digital future (Technology Magazine, 17 March 2026). Telkom Indonesia puts AI inside network modernization in its TLKM30 transformation plan and links its AI units through a platform called AIcosystem (MTN Consulting profile, 9 October 2026). That shows in the revenue evidence. IOH reports AI Cloud revenue of $33 million in 1H26. Telkom has stated AI revenue once, as a figure that Industry.co.id reported as cumulative over seven years.

Figure 1: Indonesia stretched across Europe: the network Telkom and IOH run

A hard country to wire

Indonesia’s geography would test any network engineer. IOH counts 120 to 150 fiber cuts a day nationwide, much of it from ongoing construction, according to MTN Consulting’s July 2026 profile of IOH. In some areas, the power grid runs only 20 to 22 hours a day. As one IOH network executive said, “in some countries you don’t need that much resiliency but here you do.”

Prices don’t leave much room either. IOH’s mobile average revenue per user (ARPU) sits around $2.70 a month, against roughly $4 to $6 in comparable emerging markets, according to the same profile. That leaves less revenue per subscriber to pay for the network than in those markets (see Figure 2).

Figure 2: Two of the four hard conditions, fiber cuts and ARPU, are IOH’s own figures

Subscribe now

Both telcos beat the group average margin, but Telkom’s lead fell from 17.2 to 7.4 points

Hard conditions haven’t stopped either company from making money. Telkom’s operating (EBIT) margin was 23.3% in the four quarters to 2Q26, against a group average of 15.9% (78 telcos), according to MTN Consulting’s Global Telco Market Tracker. The group average is total EBIT divided by total revenue. Telkom beat it in every trailing four-quarter period from 4Q11 to 2Q26, but its lead fell from 17.2 points in 2Q22 to 7.4 in 2Q26, with a low of 7.0 in 1Q26. Telkom’s margin fell 8.8 points over that span while the group average rose 1.0. Its earnings before interest, taxes, depreciation and amortization (EBITDA) margin was essentially unchanged over the latest year, at 49.4% in 1H26 against 49.5% in 1H25, according to MTN Consulting’s profile of 9 October 2026. The profile estimates that part of the EBIT margin fall sits in depreciation and amortization. IOH posted a 19.9% EBIT margin in FY2025, up from 19.4% in FY2024, against a group average of 16.0% in calendar 2025, according to MTN Consulting’s July 2026 profile of IOH (see Figure 3).

Figure 3: Telkom’s lead over the group average (78 telcos) fell from 17.2 points to 7.4

Both also keep building. In each of the 15 quarters from 4Q22 to 2Q26, IOH and Telkom spent a larger share of revenue on capex (capital intensity) than the group average (119 telcos). IOH’s margin over it fell to 0.4 points in 2Q23, and Telkom’s fell to 1.0 points in 1Q26. Telkom spent more than 20% of revenue on capex in every calendar year from 2013 to 2023, between 21.3% in 2021 and 29.1% in 2014. Its capex then fell from 24.2% of revenue in 2022 to 17.1% in the four quarters to 2Q26, against a group average of 16.1%. IOH put 23.0% of its revenue back into the network in the four quarters to 2Q26, down from 28.9% a year earlier (see Figure 4). IOH’s 9 June 2026 release announced a Nokia modernization of its radio network, with mid-band 5G across about 80% of the network over three and a half years.

Figure 4: IOH and Telkom capital intensity consistently above the global average

IOH’s 26.9% headcount rise is at most 9.1% once the 2022 merger is counted, while Telkom cut 16.6%

According to MTN Consulting’s Global Telco Market Tracker and Telco Workforce Tracker, IOH’s headcount rose from 3,353 in 2Q22 to 4,256 in 2Q26. Telkom’s fell from 20,897 to 17,438. The median across 72 telcos fell 6.9%. IOH’s rise starts from a base that understates the merged company. The merger of Indosat and Hutchison 3 Indonesia took effect on 4 January 2022 (CK Hutchison release), and the combined company reported more than 3,900 employees (tfr.news, 19 February 2022). From 3,900, the rise to 2Q26 is at most 9.1%. The tracker estimates interim quarters, and its 2022 quarters each add exactly 231 employees, which spreads the merger’s one-time step across the year. From the 2022 year-end count of 3,815 to the 2025 year-end count of 4,202, the rise is 10.1%.

Revenue per employee puts IOH well above the median, and most of its decline since 2Q22 traces to the same headcount base. IOH generated $845,000 of revenue per employee in the four quarters to 2Q26, 1.76 times the 72-telco median of $481,000. That ranks 20th of the 118 telcos in the Global Telco Market Tracker. Telkom, at $467,000, is 3.0% below the median. IOH’s revenue per employee fell 12.2%, from $963,000 in 2Q22 to $845,000 in 2Q26, or 3.2% a year, while the median rose 4.4% a year. Telkom’s fell 0.9% a year, from $484,000 to $467,000. With 3,900 employees in 2Q22 instead of 3,353, IOH’s annualized second-quarter revenue per employee would be $821,000 in 2Q22 and $827,000 in 2Q26 (see Figure 5). The window also includes the 1H26 transfer of management of more than 86,000 km of fiber to a fiber joint venture, FiberCo, agreed on 23 December 2025 (IOH 1H26 release).

Figure 5: IOH’s revenue per employee is 1.76 times the median, but its headcount rise starts from a base below the merged company’s size

Telkom attributes its smaller workforce to organizational streamlining and an Early Retirement Program (ERP), which began in December 2025 with 612 participants, and it has not linked any workforce change to AI (MTN Consulting TAIA profile, 9 October 2026). Both telcos also invest in AI skills. Huawei says its AI operations work with IOH has upskilled more than 400 operations staff. Telkom launched an AI Center of Excellence in September 2025. Its AI Campus partnership with Universitas Muhammadiyah Yogyakarta aims at 113 thousand digital talents, with no stated deadline. Telkom Corporate University signed a memorandum with Universitas Gadjah Mada on 15 November 2025 covering Center of Excellence research in energy and health. Readers who want to compare these workforce trends with 145 other telcos can find them in MTN Consulting’s Telco Workforce Monitor.

Five of eight featured AI results measure network operations, and only one measures revenue

The two TAIA profiles list 38 deployments, 15 at IOH and 23 at Telkom, with statuses that run from trial and announced to in production. Most of the results reported for them come from the telcos themselves or from vendors that sell them technology. Of the eight results featured here, five measure network operations, two measure customer service and sales, and one measures revenue, IOH’s $33 million of AI Cloud revenue in 1H26 (see Figure 6).

Read more

Bottom Line Up Front

This issue covers 15 September to 14 October 2026, a window in which telcos and their vendors reported AI results, AI-RAN trials and workforce changes.

Three items carry the most weight this issue. First, a TM Forum survey shows that telcos plan to raise autonomous network spending before most can show a financial return. Second, Deutsche Telekom (DT) set a €1.1 billion AI and automation savings target for 2027. Its finance chiefs say wage inflation and AI running costs offset that gross figure. Third, Nokia’s 2028 AI-RAN target is at least double what a study it endorsed projects.

MTN Consulting’s Gen AI Telco is tasked with tracking telecom’s AI transition, and basing analysis on credible, audited data. That is a challenge, because almost every data point released about AI comes from someone who has something to gain. Vendors release studies saying their AI solution cut costs. For this edition, we found that 42 of the 57 items logged came from a person or organization with an incentive to exaggerate. This is not unique to AI but it’s important to note. This market is young and information is very imperfect.

Telcos also struggle to show returns on autonomous networks. In a TM Forum interim survey of 78 telcos, 77% plan to raise autonomous network investment. Only 62% can link it to operational benefits and 37% can link it to financial outcomes.

Proof is also thin on the vendor side. AI-RAN builds AI into the radio access network (RAN), the base stations and antennas that connect phones to the network. The AI-RAN Alliance, an industry group launched in 2024, works in three areas. They are AI that improves the RAN, AI applications that run on it, and shared infrastructure for both.

Nokia says it has demonstrated a 20% gain in spectral efficiency, the amount of data a radio carries per unit of spectrum, and targets more than 100% by 2028. Ericsson cites up to 10% in spectral efficiency across Optus sites. Orange, which is taking Nokia’s AI-RAN into live-network testing, plans to start small.

On job cuts, AI appears in savings targets but is rarely blamed. DT targets about €1.1 billion of gross savings from AI and automation outside the US in 2027. Its chief human resources officer, Birgit Bohle, said that a task-by-task review across the company already shows where DT needs fewer resources. DT gave no job count. DT’s group headcount fell 6,283 full-time equivalents (3.2%) in 1H26, and its interim report ties 4,671 of them to T-Mobile US’s workforce transformation. Telia plans about 900 position cuts, but its release never mentions AI. Telcos have little reason to blame AI for job cuts when opposition to AI is rising in some markets, so that silence settles nothing. MTN Consulting’s Telco Workforce Tracker shows that headcount at DT, Telia, Verizon, Telefónica, SK Telecom and LG Uplus fell 9% to 29% from 2Q22 to 2Q26. The role of AI in these cuts is undetermined.

Elsewhere, spending and capacity kept growing. According to the South China Morning Post, China Mobile’s Hohhot computing hub reached 20,100 petaflops, 3 times its 2024 level. SK Telecom (SKT) told employees it will build a Token Gateway business, and etnews reports an internal SKT plan for 20 gigawatts (GW) of AI data center capacity by 2035.

Subscribe now

Signal vs. Noise

Below is a summary of recent developments in telecom AI, automation and autonomous networks, our view on why they matter, and their implications for operators, vendors and investors.

Nokia’s 2028 AI-RAN target is more than 100% spectral efficiency gain by 2028

16-Sep-26 to 12-Oct-26

Nokia says it has demonstrated a 20% spectral efficiency gain, Orange will start small in its live network and Samsung’s two Korean field projects start in October

Nokia sets the pace on AI-RAN claims, but the telcos closest to its roadmap are moving more cautiously than the vendor’s targets imply. Most of the gains below are in spectral efficiency, the amount of data a radio carries per unit of spectrum.

Ericsson’s claims rest on a rollout with few published specifics. Optus is rolling out Ericsson’s dynamic link adaptation across its Australian network, per its network chief technology officer, Sriharan Amirthalingam. Ericsson product manager Melike Erol-Kantarci put field gains at up to 10% in spectral efficiency across sites and up to 25% in certain places. Neither figure comes with a baseline or a site count. Ericsson’s March release on the Optus trial reported cell-level throughput gains of more than 20% in medium to poor radio conditions on one site, again without a baseline. Optus itself has not released results.

Nokia’s claims are larger and further out. At the Intelligent RAN Forum, Nokia said it has demonstrated a 20% spectral efficiency gain and that it targets 50% by mid-2027 and more than 100% in 2028. The report of the forum names no baseline for any of the three figures. Nokia said on 16 September that eight telcos are advancing AI-RAN proofs of concept and live trials on Nvidia platforms. They are A1 Group, Chunghwa Telecom, du, e&, Mobily, stc, TPG Telecom and Zain Saudi.

Orange, the telco closest to Nokia’s roadmap, sets a more cautious pace. Its group chief technology officer, Laurent Leboucher, says that Orange will start small in its live network and validate performance, energy use and total cost of ownership before extending. He added that Orange will not commit to one architecture and that upgrading existing AirScale equipment (from Nokia) may be the most pragmatic option in many places.

Nokia’s GPU-based RAN products already face doubt from Verizon. Nokia and Nvidia will sell graphics processing unit (GPU) based RAN products from early next year. About a year ago, Verizon chief technology officer Yago Tenorio said that telcos will not run RAN workloads on those GPUs but should use them for AI inference.

Other estimates of the gain are lower than Nokia’s target. A white paper by Chetan Sharma Consulting that Nokia endorsed projects 40% to 50% under favorable conditions. Nokia’s 2028 target is at least double the top of that range. Marcus Weldon, a former Nokia chief technology officer, puts the realistic gain at 20% to 30%, closer to Ericsson’s published figures. He calls a doubling possible only in niche scenarios.

Korea may supply the next field evidence. Samsung says it is the sole global vendor for KT and the main vendor for SKT in two government-funded Hyper AI Network projects. The projects, starting in October 2026, cover field trials at the HD Hyundai Samho shipyard and validation at SK Incheon Petrochem.

Why it matters: Buyers should treat every AI-RAN gain claim as an upwards bound target, unless the result is published and has a clear baseline. Ideally the result would also be audited by an independent third party, but that almost never happens. Ericsson sells its package as software for the installed base, with no new hardware, so its claims are cheap to test. Nokia’s larger gains depend on new hardware, and the GPU decision turns on whether a telco sees the GPU as RAN capacity or as an inference server. Verizon’s chief technology officer has already said the second. Orange’s live-network validation, the two Korean projects and the first telco comment on Nokia and Nvidia’s products will show which view holds.

Source: https://www.nokia.com/newsroom/nokia-accelerates-ai-ran-adoption-as-global-operators-embrace-ai-native-network-evolution-on-nvidia-platforms/

Source: https://www.ericsson.com/en/press-releases/7/2026/optus-and-ericsson-trial-ai-to-boost-5g-downlink

Source: https://www.lightreading.com/5g/nokia-to-start-dealing-out-gpus-in-high-stakes-ai-ran-gamble

Deutsche Telekom targets €1.1 billion of gross AI savings; its HR chief says it already sees where it needs fewer resources

05-Oct-26

Group headcount fell 6,283 full-time equivalents (3.2%) in 1H26, and the interim report ties 74% of the fall to T-Mobile US’s workforce transformation

DT announced goals for AI revenue and savings at its AI Investor Day in Bonn on 5 October 2026. The company’s press release reports about €250 million of AI-related revenue outside the US in 2026 and targets about €800 million by 2030. DT’s presentation raises its 2027 estimate of gross savings from AI and automation to about €1.1 billion, from about €0.8 billion at its 2024 capital markets day. About €1.0 billion is indirect costs and about €0.1 billion is capital expenditure. It puts targeted gross savings at about 4% of indirect costs in 2026, 6% in 2027 and 15% in 2030, or about €2.5 billion. All figures compare with a 2023 baseline, exclude the US and token costs, and measure indirect costs after leases but including capitalized labor.

The savings do not lift DT’s profit guidance. DT confirmed two 2027 targets. Adjusted earnings before interest, taxes, depreciation and amortization (EBITDA) after leases is about €16 billion, and free cash flow after leases is €3.7 billion to €3.9 billion. DT says AI-related operating savings are reinvested in accelerated digital transformation. It does not say what share it reinvests.

DT’s finance chiefs explained the flat guidance on stage, in the event transcript. Christian Illek, the current chief financial officer, said DT is investing heavily but has not yet scaled most projects. He admitted that some projects failed and ended up costing more than they saved. Each business segment pays for its own token use, and he expects token costs to stay below a low double-digit percentage of gross savings. Dhananjay Mirchandani, who becomes chief financial officer in May 2027, said it is too early to attribute margin gains to the savings. He cited wage inflation and the running cost of AI systems as offsets. The €1.1 billion is therefore a gross figure, and DT has not netted it against those costs.

Sales and service carries the largest 2030 savings target of any function on DT’s slides. DT also aims for AI to handle every initial call and for calls and chats to fall 50% to 55% from 2023. The release says its Frag Magenta chatbot handled about 2.6 million calls in 1H26, US customer service calls fell 55%, and AI agents handle 40% of US customer contacts. A T-Mobile US executive told the event that calls handled by humans in customer care were down 55% in 2Q26. The target is 75% by the end of 2027. T-Mobile US’s slide shows annual contacts per account falling from 7.0 in 4Q21 to 3.1 in 2Q26, a 56% fall, which may be the same measure. That slide puts the share of postpaid contacts served by an AI agent at 32%, against 40% in the release, but the two may not measure the same thing.

The workforce slides give no numbers. DT’s workforce redesign slide shows the current workforce mix evolving in three possible ways: reskill, build up, or reduce. It lists an assessment of AI potential by role, task and activity as the method, and the operating-model slide says impact and workforce effects stay transparent. The presentation contains no headcount target, but it does share an example. For a midmarket logistics firm with 100 employees, DT says its agent platform releases 15 to 20 full-time equivalents worth €0.9 million to €1.2 million a year. It gives no such estimate for DT’s own workforce.

The executives said more about jobs on stage than the slides do. Birgit Bohle, DT’s chief human resources officer, said no one knows what AI will mean for the workforce over ten years. She also said DT’s task-level review of roles already shows where it needs fewer resources, and that DT will accelerate its workforce transformation in a socially responsible way. Chief executive Tim Höttges described an expense claim that passes through 10 to 12 departments and said an agent can cut people out of that chain. Group technology officer Arash Ashouriha said the 30% network productivity target could mean handling more work or moving network experts to other tasks.

DT’s own interim report breaks down where headcount went. Group full-time equivalents fell from 198,079 at the end of 2025 to 191,796 on 30 June 2026, a drop of 6,283, or 3.2%. T-Mobile US accounts for 4,671 of that, or 74%, which the report attributes to its 2025-2026 workforce transformation. Germany fell 928, helped by phased retirement, and Europe fell 443, mainly in Greece and Poland. The report names no AI cause for any of these declines. T-Mobile US also ties its cuts to retail store closures and the UScellular acquisition. The MTN Consulting Telco Workforce Tracker puts DT’s headcount at 210,595 at 2Q22 and 191,796 at 2Q26, down 8.9%. The fall came late: headcount was flat in 2025 but fell 3.2% in 1H26.

AI evidence from the network is more focused than for customer-service. DT’s presentation says its RAN Guardian agent made incident handling more than 95% faster, from one hour to one minute. The agent monitors and corrects congestion in the radio access network at large events, and DT says it widened event coverage from 1,000 to 100,000 events. DT’s slide gives one hour as the old handling time, so one minute is a 98.3% reduction, consistent with the “more than 95%” claim. MINDR is DT’s multi-agent system for faults across radio, transport and core, but it is not yet in production. So, its slide lists targets, not results: more than 90% of tickets handled automatically and network restored within 60 minutes of any incident. Ashouriha called MINDR a “minimum viable product” that DT is still hardening. DT puts network gross savings at 10% to 15% by 2030. In 2025, outside the US, DT’s network operating cost was €4.3 billion and its capital expenditure was €5.8 billion.

Light Reading estimates, by a calculation it calls crude, that up to 25,000 roles could be at risk. The estimate assumes all savings come from headcount cuts spread evenly across salary bands, against about €19.8 billion of personnel costs last year. DT’s slides do not show how much of the savings is payroll.

The 1H26 decline shows no AI effect. Most of it sits in a US unit that cites restructuring and store closures, and the German decline came through phased retirement. T-Mobile US also moved software engineering from an outsourced workforce to its own staff, which raises an employee count. DT’s targets point to a different outcome from 2027. DT attributes its savings to AI and automation and expects bots to take first contact in customer service. Its human resources chief says DT already sees where it needs fewer resources. A headcount effect is plausible from 2027 and unproven today.

Why it matters: Investors can hold DT to the €1.1 billion for 2027, but they cannot use the 1H26 headcount decline as evidence for it. DT’s finance chiefs say wage inflation, AI running costs and the cost of failed projects offset the gross savings, so the figure is not a margin forecast. Network vendors such as Ericsson and Nokia face a buyer that builds its own agents. DT’s group technology officer, Arash Ashouriha, said that DT’s agents never act on a network element directly. They act through a Model Context Protocol (MCP) gateway that limits what each agent may do. DT developed RAN Guardian and MINDR in-house. Its chief network officer for Europe, Armin Sumesgutner, said DT starts from what the industry offers and builds on top of it. Vendors that publish interfaces through which an agent can query network state and perform permitted actions supply what DT’s gateway design needs. Vendors that sell closed agents of their own compete with the systems DT has already built.

Source: https://www.telekom.com/en/newsroom/latest-updates/media-information/2026/10/deutsche-telekom-boosts-growth-efficiency-and-quality-through-t

Source: https://www.telekom.com/en/investor-relations/publications/capital-markets-days/ai-investor-day

Source: https://report.telekom.com/interim-report-q2-2026/management-report/development-of-business-in-the-group/non-financial-metrics-of-the-group.html

Globe credits AI with PHP1.2 billion of value, 68% of it incremental revenue

25-Sep-26

A 2.8x return on AI investment implies about PHP430 million of spend if it divides value by investment

Globe Telecom split its AI value into revenue, cost and time. In a recent release, Globe reports that its AI programs produced PHP1.2 billion of measurable value as of August 2026. This is comprised of over PHP825 million of incremental revenue, PHP368 million of cost avoidance and PHP28 million of productivity value, which the release equates to over 52,000 labor hours saved. Globe also reports a 2.8x return on AI investment and 90% enterprise AI adoption, and says its community of AI builders has grown to over 450 employees.

Globe’s claims require a caveat. Incremental revenue needs a counterfactual, meaning an estimate of what revenue would have been without AI, and the press release gives none. Cost avoidance counts spending that did not happen, so it cannot be found in the income statement. The release does, however, list the areas addressed by AI at Globe: finance, customer engagement, marketing technology and order management. The productivity line values 52,000 hours at PHP28 million, or about PHP538 an hour. At 2,080 working hours a year, those hours equal about 25 full-time employees for one year. Globe does not say whether the hours reduced anyone’s workload or headcount.

The 2.8x return has no stated formula. If it divides value by investment, Globe’s AI investment would be about PHP430 million (PHP1,200 million divided by 2.8). The release does not give the investment, so the figure is implied, not reported. Globe’s chief executive, Carl Cruz, says the real measure of AI is what it improves, not how many people use it.

Why it matters: Globe is the only telco in this issue that splits realized AI value into revenue, cost and time. Other telcos may be asked to match that template. Investors and AI vendors should ask for assumptions behind the revenue counterfactual and the investment base behind the 2.8x. They should also ask whether the avoided costs repeat every year or occur once.

Source: https://www.globe.com.ph/about-us/newsroom/corporate/globe-ai-adoption-impact-1-2b-value

Telstra’s AI token use grew 275%; 80% goes to customer service and code

22-Sep-26 to 06-Oct-26

Its technology chief rates Telstra’s network below Level 2 on TM Forum’s autonomy scale of 0 to 5 and tells telcos to fix basics first

Telstra shows where a telco’s AI actually runs. At FutureNet Asia, Group Executive Kim Krogh Andersen said Telstra’s token consumption grew 275% last year. It now averages 15 billion tokens a day across more than 400 use cases. He said costs fell 7%. He gave the token usage split: customer experience 56%, software engineering 24%, autonomous networks 12% and cybersecurity 8%. Customer experience and software engineering together take 80%.

Andersen did not say whether the 7% is total cost or cost per token. Most likely it is total cost. That would imply that token volume rose 3.75 times while spend fell 7%, which implies a cost per token about 75% lower.

On network autonomy, Andersen said a recent internal assessment places Telstra below Level 2 on TM Forum’s autonomous networks scale, which runs from Level 0 to Level 5. He said the industry will not reach Level 4 without retiring legacy systems.

Telstra’s production agents also work in the contact center. Salesforce says two Agentforce agents are in production at Telstra and together support more than 1,000 contact center representatives. One guides refunds during fixed-service outages and one audits access permissions. Salesforce expects such agents to resolve hundreds of thousands of routine inquiries a year.

Telstra’s headcount has fallen since 2023. In the MTN Consulting Telco Workforce Tracker (2Q26 edition), Telstra’s year-end headcount peaked at 32,579 in 2023. It stood at 29,334 at 2Q26, a fall of 10.0%. The longer comparison looks different. From 2Q22 to 2Q26, Telstra’s headcount rose 1.5%, from 28,889 to 29,334, while the median across 72 telcos fell 6.9%. That base sits just after a 4,906 jump in Telstra’s 2022 year-end count. Telstra added about 1,700 Digicel Pacific staff on 14 July 2022, which accounts for part of that jump, and the tracker does not adjust for acquisitions or divestitures.

Revenue per employee fell 1.5% a year at Telstra against a median rise of 4.4%. Telstra’s earnings before interest and taxes (EBIT) margin rose 10.2 percentage points against a median rise of 0.1. MTN Consulting’s profile records a statement by chief executive Vicki Brady at the 27 May 2025 investor day. She said she believes Telstra’s workforce will likely be smaller in 2030 than today.

Why it matters: Telstra’s token mix shows that its AI use concentrates in customer experience and code, which take 80%, while autonomous networks take 12%. Telstra’s technology chief names retiring legacy systems as the precondition for network autonomy, so vendors selling autonomy tools should expect that work to come first. Vendors should concentrate marketing energy on helping telcos retire this legacy infrastructure, not just encourage them to deploy the next-generation.

Source: https://www.lightreading.com/ai-machine-learning/telstra-s-tech-chief-asks-telcos-to-fix-basics-before-chasing-ai

Source: https://salesforce.com/au/news/press-releases/2026/09/22/telstra-uses-salesforce-agentforce-to-improve-customer-experience

Source: https://www.telstra.com.au/exchange/expanding-the-telstra-family-with-digicel-pacific

China Mobile’s Hohhot hub tripled to 20,100 petaflops as China Telecom’s token platform passed 170 models

23-Sep-26 to 13-Oct-26

Both numbers measure supply: a petaflop figure is capacity and a model count is catalog size

China’s state telcos have been talking about network capacity recently. For example, China Telecom general manager Liu Guiqing spoke at the 2026 China International Information and Communications Technology Exhibition on 23 September. He said the company manages more than 118 exaflops of computing power across 16 data center parks of the 100-megawatt class. He said its Xingchen Token Hub now supports more than 170 large models. He added that China Telecom will form an AI Token global service ecosystem alliance. Our previous issue reported 142 models on the platform from China Telecom’s interim results, so the count rose by at least 28, or about 20%, since August.

China Mobile’s capacity news concerns a single hub. Citing Inner Mongolia Daily, the South China Morning Post reports that the Hohhot hub reached 20,100 petaflops after an expansion. Of those, 16,700 petaflops serve AI and smart computing. A 2024 state press report put the hub at 6,700 petaflops, so capacity tripled. The newspaper does not say whether the two figures use the same numerical precision, so the tripling may not be like for like.

Table B below shows the procurement behind this capacity. It includes a China Mobile tender for 20,480 AI cards and C114’s count of 36.6 billion yuan of telco data center tenders in 3Q26.

Why it matters: A model count measures catalog size, not usage, and a petaflop figure measures capacity, not demand. Token volumes and revenue per token would show whether selling tokens by subscription works. A token revenue line in any Chinese state telco’s 3Q26 results would answer that.

Source: https://www.c114.com.cn/news/117/a1318271.html

Source: https://www.scmp.com/tech/big-tech/article/3370600/china-mobile-triples-capacity-mega-inner-mongolia-computing-hub-amid-nations-ai-push

SK Telecom runs 14 million AI requests a day on Rebellions chips and plans a Token Gateway business

21-Sep-26 to 12-Oct-26

etnews reports an internal SKT blueprint for 20 GW of AI data center capacity by 2035, citing industry sources

Rebellions, the Korean AI chip maker, reports that its ATOM-Max servers handle about 14 million real-time requests a day across four SKT services. They process more than 4 billion tokens a day. The servers are built on neural processing units (NPUs). The first of the four, a call summary feature in the A. dot app, entered commercial use in December 2025. A streaming text-to-speech service, an AI call center post-consultation and search system, and a scam bait-text detector followed from June 2026. The figures come from Rebellions, which merged with SAPEON Korea, a unit linked to SKT, in December 2024.

SKT’s chief executive, Jung Jae-heon, presented a Growth Gear strategy at an employee town hall on 8 October. It links telecom, AI services and AI infrastructure around a goal of leading the token economy. SKT’s Token Gateway business will route enterprises and customers to the best-suited model or infrastructure. It builds on SKT’s roughly 23 million customers, its payment system and its AI platform. Jung said AI transformation has already improved efficiency by automating repetitive work, and that the next step is to redesign work methods and organization around it. SKT will run an internal open call soon to move employees into core growth businesses. SKT’s release targets commercial service of its AI factory with Nvidia for next year.

etnews adds a larger number for the infrastructure side. Citing telecom industry sources, it reports that Jeong Seok-geun, head of SKT’s AI unit, shared an internal blueprint for 20 GW of AI data center capacity by 2035. The plan combines the 15 GW SKT has already announced for Korea and 5 GW overseas, likely with local partners. etnews says SK Group chairman Chey Tae-won named Japan as the first overseas target in a Nikkei interview.

The open call is SKT’s workforce news this month. SKT frames redesigning work around AI and moving staff into growth businesses as redeployment, and it gives no position numbers.

Why it matters: SKT is moving toward selling tokens and compute, but its public evidence so far is request volume from a chip supplier, not revenue. The Token Gateway puts SKT beside the Chinese token platforms above and the Singtel and Indosat services in Table C: a metered AI product with a per-unit price. For chip vendors, Rebellions shows that a Korean NPU can carry production telco traffic. For investors, the first disclosed Token Gateway revenue is the test.

Source: https://kr.rebellions.ai/newsroom/rebellions_running_14000traffic_skt_commercialization/

Source: https://news.sktelecom.com/231883

Source: https://www.etnews.com/20261012000272

77% of surveyed telcos plan to raise autonomous network spending, but only 37% can show the financial return

07-Oct-26 to 08-Oct-26

TM Forum says 72% of telcos believe their AI is trustworthy but only 14% can verify it

TM Forum’s interim survey shows how few telcos can link autonomous network spending to results. The survey of 78 telcos finds that 77% plan to raise autonomous network investment, 62% can link it to operational benefits but only 37% can link it to financial outcomes. A further 51% say vendors provide too little tooling to measure return. Final results are due in December 2026.

A second TM Forum finding concerns trust. In its announcement of a Trustworthy AI program with Accenture, TM Forum says that 72% of telcos believe their AI is trustworthy but only 14% can verify it. The figures come from “recent TM Forum research” that the release does not describe or link. The two organizations will build an AI trust certification over several years, with no launch date.

Broadband providers also cite data and AI integration as obstacles. In a Broadband Forum survey of 100 broadband network decision makers, 94% plan to use software-defined networking and 92% call AI and machine learning integration critical or important. In the same survey, 83% cite poor, inconsistent data as a major barrier. STL Partners produced the report, and Radisys and Incognito Software Systems sponsored it. Both sponsors sell software to broadband providers.

Why it matters: Vendors selling autonomous network tools face buyers with budget intent but no internal proof of return. That gap lengthens sales cycles and favors vendors that ship measurement with the product. The 51% who say vendors provide too little tooling to measure return point to demand for built-in measurement. A telco that publishes a result with a baseline and a financial outcome stands apart from the 63% who cannot.

Source: https://inform.tmforum.org/features-and-opinion/an-has-a-benchmarking-challenge-the-road-to-roi-must-be-measured

Source: https://www.tmforum.org/news-insight/newsroom/tm-forum-launches-industry-initiative-to-enable-trusted-ai-autonomy

Source: https://www.broadband-forum.org/news/2026-10-08-broadband-service-providers-identify-major-hurdles-in-the-transition-to-autonomous-networks-new-report-finds/

LG Uplus cut average contact center waits from 47 seconds to 5

15-Sep-26 to 08-Oct-26

Another eight telcos report volumes and speed gains

A growing list of telcos is opening up about their AI deployments, sharing data on volumes and speeds. LG Uplus is unique among recent reporters, as it reports clear before and after results, but it measures wait time, not cost. Verizon’s “thousands” of saved technician hours and T-Mobile’s “about half the time” give no base against which to test them.

Why it matters: Buyers can compare these deployments on scale and speed, although not on financial return. A telco that publishes a saving with a baseline would give buyers a comparison point.

Source: https://news.lguplus.com/22900

Source: https://www.t-mobile.com/news/network/t-mobile-ai-and-resilience-make-5g-network-stronger

Source: https://rcrwireless.com/20261001/ai-infrastructure/agentic-verizon-draws-line-marks-difference

Chinese telcos tendered 36.6 billion yuan of data center infrastructure in 3Q26; Deutsche Telekom’s incoming chief financial officer indicates no extra radio spending for AI

30-Sep-26 to 13-Oct-26

DT guides 2027 capex at 21% of service revenue excluding the US

Spending news related to telco use of AI helps understand priorities. Deutsche Telekom’s incoming chief financial officer could recall no discussion of extra radio spending for AI, so on his account AI-RAN has not changed DT’s plan. By contrast, Chinese telcos are carving out big additional chunks of budget for AI factories. For instance, China Mobile recently floated a tender for over 20K AI cards (GPUs). That’s separate from the big recent tender for civil works related to the AI buildout in China: per C114, the 36.588 billion yuan total spend counts only power, cooling, civil works and cabling.

Why it matters: The two buyers in this story draw on different vendors, so the signals matter to different readers. For Chinese AI chip and server makers, China Mobile’s September notice seeks 20,480 AI cards, 3.3 times the 6,208 cards it tendered in March. The March tender specified Huawei’s Compute Architecture for Neural Networks (CANN), the software platform for its Ascend AI chips. C114 says that order shows domestic computing power can handle national-scale AI tasks. The September notice names no chip, so it does not show whether domestic suppliers win the larger order. Chinese telcos do have a reason to favor them, because export controls limit access to advanced foreign AI chips. For vendors selling radio equipment to Western telcos, DT’s incoming chief financial officer recalls no extra radio spending for AI. On this evidence, vendors should not expect AI to lift DT’s radio orders.

Source: https://www.lightreading.com/ai-machine-learning/deutsche-telekom-dashes-vendor-hopes-for-big-mobile-uplink-spending

Source: https://www.c114.com.cn/news/118/a1318698.html

Source: https://www.c114.com.cn/news/16/a1319018.html

Source: https://m.c114.com.cn/w118-1307117.html

Smart priced its AI service from PHP49 for 7 days, the one published price among eight telco AI launches

22-Sep-26 to 09-Oct-26

Recent launches take three forms: metered tokens, cloud capacity and partnerships

Telcos announced three kinds of AI service this month: metered tokens (Singtel, Indosat, Smart), capacity and cloud (Uztelecom, Claro), and partnerships (Bell twice, VEON). Smart is the only one to publish prices, but as a prepaid consumer product it says little about enterprise demand. Indosat names a target of 50 businesses at launch, but that’s just a goal. VEON’s 27% share covers digital and AI services together, so it cannot size AI revenue.

Why it matters: Telcos have launched token and GPU services, but none has shown that anyone is buying them. Seven of the eight launches publish no price, and none discloses a customer count or an AI revenue figure. Indosat’s target of 50 businesses is the only customer figure, and it is a goal. Smart’s price list, from PHP49 for 7 days to PHP499 for 30 days, covers a consumer prepaid product and tells enterprise vendors little about business pricing. Vendors selling metering, billing or GPU capacity to these telcos should ask for the demand case behind each launch before they commit capacity. Investors should treat the first disclosed token revenue line, even a small one, as the first evidence that the category sells.

Source: https://www.singtel.com/about-us/media-centre/news-releases/singtel-reai-launches-ai-token-as-a-service

Source: https://www.f5.com/company/news/press-releases/indosat-and-f5-launch-intelligence-suite-a-token-factory-to-unlock-ai-across-asia

Source: https://tribune.net.ph/2026/09/22/smart-targets-prepaid-market-with-low-cost-ai

72% of surveyed operators plan to use digital twins but only 13% do today; 8% of UK online adults use AI “in relation to” their telecoms services

23-Sep-26 to 13-Oct-26

Other recent news items center around plans, claims or a committee launch, and BT’s £3.67 billion estimate comes from the seller of the product it supports

Standards bodies and surveys supplied the month’s autonomy benchmarks, but most of them describe plans. TM Forum’s survey of 128 operators shows the gap: 72% plan to use digital twins to enable autonomous networks and agentic AI, but only 13% already do. Ofcom’s survey is the exception: it asks consumers what they used. It finds that 8% of online adults use an AI tool in relation to their telecoms services. The uses are resolving an issue (6%), comparing prices, products or services (3%) and making a complaint (2%). Those uses overlap, since the shares sum to 11%. AT&T’s Level 4 RAN claim has no outside assessment behind it, and TM Forum’s 20 solution packages name no adopter. BT claims its new AI Receptionist tool helps customers recover missed revenues, but this is based purely on a BT-sponsored model.

Why it matters: Operator plans run well ahead of use. A digital twin is a virtual model of a network used to test changes before they go live, and the survey finds far more operators planning one than running one. Consumer use is narrower, and its largest category is resolving an issue. That is a customer-care task, not network autonomy. Vendors selling autonomy tools are selling into plans, so they should expect buyers to ask for a named adopter with a measured result. Telcos choosing where to start with customer-facing AI have one measured use to build on: resolving customer issues.

Source: https://www.ofcom.org.uk/phones-and-broadband/telecoms-infrastructure/consumers-using-ai-to-help-manage-their-telecoms-services-but-human-contact-still-valued

Source: https://www.tmforum.org/news-insight/newsroom/tm-forum-launches-reusable-blueprints-to-accelerate-level-4-autonomous-networks

Source: https://newsroom.bt.com/ai-from-bt-launched-to-help-uk-firms-adopt-ai-with-confidence-targeting-37-billion-lost-through-missed-calls/

ZTE and XLSmart’s agent reached a 25% closed-loop completion rate in a live network test

10-Sep-26 to 08-Oct-26

Ericsson quotes 4x capacity against its own earlier processor, a vendor comparison

The most useful evidence on the vaue of telco AI would come from a telco directly, and be based upon a live network test. Of the 7 items in this story, only one fits this bill. XLSmart and ZTE tested a network optimization agent on live coverage-optimization scenarios; the closed-loop completion rate was 25% within the tested scope. The article, by Selular.ID, says the agent analyzes network conditions, drafts an optimization strategy and executes the needed actions. The article gives no scenario count. The article presents the test as one stage toward Autonomous Network Level 4 and says the results show which parts still need human intervention: about 75% of cases.

Ericsson also announced deployment of a new RAN processor at Vodafone Egypt. Ericsson says its 4x capacity and 50% energy figures compare its new processor with its own earlier generation. The other five of the seven items quote no performance figure.

Why it matters: A 25% closed-loop completion rate is low for a test that the article presents as one stage toward Level 4, where a network increasingly runs itself. It is the only figure in this group from a live test with a telco. Telcos evaluating autonomy agents should ask each vendor for the completion rate in a live network test, the number of scenarios tested and how the unfinished cases were handled. The answers separate a tested product from a launch announcement.

Source: https://www.ericsson.com/en/press-releases/1/2026/vodafone-egypt-brings-ai-with-ericsson-processor

Source: https://selular.id/2026/09/xlsmart-dan-zte-uji-ai-agent-untuk-autonomous-network-level-4/

Source: https://www.nokia.com/newsroom/nokia-improves-network-performance-and-efficiency-for-zain-ksa-across-saudi-arabia/

Headcount fell 9% to 29% at six telcos from 2Q22 to 2Q26; Telia cites localization, not AI, for its latest 900 cuts

23-Sep-26 to 12-Oct-26

Telia, Verizon and Telefónica fell 18% to 29%, but their stated reasons name restructuring and localization, not AI

A telco’s stated reason for a job cut is something you should take with a grain of salt. MTN Consulting tests each stated reason four ways. The tests ask whether the reason names a function that AI automates, and whether the company claims AI savings in that function. They also ask whether a non-AI explanation exists, and whether the company made similar cuts before without citing AI.

Telia illustrates why the tests matter. Its 900 announced cuts are the only new job-cut number among the nine telcos in the table below. Telia has now announced about 4,500 across three rounds since December 2024, but none of the three cites AI. Deutsche Telekom shows the opposite risk. It cites AI for 2027 savings but attributes its 1H26 headcount decline to other causes. The future effect is plausible but the past effect is not shown.

Per the MTN Consulting Telco Workforce Tracker, from 2Q22 to 2Q26, headcount fell at all six telcos in the table below, by 9% to 29%. Telia, Verizon and Telefónica fell 18% to 29%, but none of the three attributes the fall to AI. Verizon’s figure includes Frontier Communications staff added in 2026, so its fall excluding Frontier is larger. The tracker does not adjust for acquisitions or divestitures. Telefónica has agreed to sell its Argentine, Colombian and Peruvian businesses, so portfolio change explains part of its fall.

The tracker’s wider findings argue against attributing these falls to AI. In its 2Q26 findings, MTN Consulting reports that headcount across its 145-telco panel fell 2.1% year on year to 4,285,701. The decline began several years before the AI cycle. Across 72 telcos, a telco’s headcount change had no measurable relationship to its change in earnings before interest and taxes margin. A fall in headcount at one telco therefore needs company-level evidence before anyone credits AI for it.

Why it matters: Headcount at all six tracked telcos fell over the four years to 2Q26. The decline across MTN Consulting’s panel began several years before the AI cycle, and the telcos give restructuring, localization and retirement as reasons. A falling headcount therefore does not validate an AI product. Vendors that sell AI as a labor-cost saving should expect buyers to ask for the saving measured in the function the software touches, such as handle time per contact. Company-wide headcount cannot isolate that effect. Investors have one dated test: DT targets €1.1 billion of AI savings in 2027, so its headcount in the affected functions that year tests the claim. Regulators and employee representatives who want to know whether AI drives job cuts should ask each telco for the function affected and the headcount series for it.

Headcount series for each tracked telco in this table are in the Headcount trends and Telco detail tabs of the Telco Workforce Monitor.

Source: https://news.cision.com/telia-company/r/telia-proposes-further-localization-of-operations-and-targets-sek-1-billion-in-annual-gross-savings,c4407089

Source: https://report.telekom.com/interim-report-q2-2026/management-report/development-of-business-in-the-group/non-financial-metrics-of-the-group.html

Source: https://www.mtn-c.com/product/telco-workforce-tracker-2q26-headcount-still-falling-by-2-per-year-even-as-telcos-accelerate-ai-efforts/

Source: https://www.telecomtv.com/content/access-evolution/telef-nica-offloads-another-latam-unit-52823/

Explore the data behind the workforce table in MTN Consulting’s Telco Workforce Monitor, which tracks headcount and labor costs at 145 telcos. Cost: $799 per user seat.

Open the Telco Workforce Monitor

Share

Questions about this post? Contact the author: Matt Walker, Chief Analyst of MTN Consulting (matt@mtn-c.com)

Subscribe now

Three years on, BT expects the cuts to land at the top of its range, 50,000 to 55,000. Most have little to do with AI. BT’s chief financial officer, Simon Lowth, said the highest reductions in direct labor were in Openreach, its network arm, and Corporate Functions. In Openreach, BT ties the cut to fewer faults, which follow from the well-known lower maintenance requirements of full-fiber access networks.

“Faults across fibre and copper reduced by a record 18%. This helped us cut direct labour by more than 10%.” -CEO Allison Kirkby, on FY26 earnings call.

BT has cut 22,000 people since 2023 while leaning more on sub-contractors

BT counts its workforce as “total labour resource”: its own employees plus sub-contractors. That figure stood at 130,000 when BT set its target in its FY2023 results (BT’s fiscal year ends 31 March). The goal was 75,000 to 90,000 by FY2028-FY2030. By March 2026 it was 108,000. That’s 22,000 fewer, or 17%.

The split between employees and contractors matters. Between FY2024 and FY2026, BT’s direct labor fell from 92,000 to 77,000 while sub-contractors rose from 28,000 to 31,000. Sub-contractors now make up 29% of the workforce, up from 23%. In its FY2025 call, CFO Simon Lowth said the 6,400 drop in employees was “partially offset by a 2,400 increase in lower unit cost and more flexible subcontractors.” MTN Consulting’s Telco Workforce Monitor counts employees only, so it follows BT’s direct labor and not its total labor resource: at the end of March 2026 it shows 77,152, against 77,000 from BT. From 2Q22 to 2Q26, it shows BT’s headcount down 24.1%, from 99,087 to 75,226, against a median fall of 6.9% across 72 telcos.

Openreach is where jobs disappeared, and fiber is the cause

BT’s executives say plainly where the cuts came from. In the FY2026 presentation transcript, CEO Allison Kirkby said faults across fiber and copper fell “a record 18%,” which “helped us cut direct labour by more than 10%” at Openreach. Lowth said direct labor fell in every unit, “with the highest in Openreach and Corporate Functions.” Fiber has “60% fewer faults than copper,” Kirkby said. Fewer faults means fewer engineers sent out to fix them.

The pattern was the same a year and a half earlier. In November 2024, Kirkby said most of the reduction “was in Openreach and Business, thanks to fewer faults on fibre and continued transformation.” She added a second reason: “the smart use of technology.” Lowth said Openreach had “around five thousand fewer FTE” (full-time equivalents) over the previous 12 months.

The next round follows the same logic. BT narrowed its FY2030 target in May 2026 to the low end of the range, 75,000 to 80,000, another 28,000 to 33,000 people. Lowth listed four areas for getting there. The first, he said, is “Openreach workforce reshaping post the fibre build ramp down.” BT expects repair volumes to fall from 2.4 million in FY2026 to 1.7-2.1 million by FY2030 as customers move off copper.

BT talks about AI savings but hasn’t put a number on them

The point is not that AI has no impact on job cuts. It does, and that may grow over time. The point is that these things are hard to measure, and many things are changing simultaneously – such as the completion of fiber-to-the-home builds. It’s been widely understood for at least a couple of decades that a fully fiberized access plant would bring with it lower staffing requirements. Jansen said as much on a 2023 investor call: “When we stop building the network we won’t need that workforce.”

BT’s executives still lean on AI when they talk about the future workforce. Kirkby told the Financial Times in June 2025 that earlier headcount plans “did not reflect the full potential of AI” and that BT might be “even smaller” by 2030. In May 2026 she said AI’s “early results are encouraging,” and pointed to call centers where AI helps advisers serve customers and cross-sell. She called AI “further upside” for a simpler business.

None of that came with a figure. Across BT’s FY2026 results presentation and transcript, no saving, call reduction or headcount number is attributed specifically to AI, including the £580 million of gross annualized savings it reports for FY2026. It’s probably not even possible. The closest BT has come was in November 2024, when Kirkby said “digital transformation also delivered tens of millions in savings” in the half, alongside fewer software applications and an AI chatbot. Kirkby gave network decommissioning the same “tens of millions.” BT reported £433 million of gross annualized savings for that half, meaning cost cuts counted at a full-year rate. If “tens of millions” means under £100 million, digital transformation, which includes the chatbot, supplied under a quarter of it.

The timing does favor AI. Eleven months after Kirkby said AI might make BT “even smaller,” BT cut the top of its FY2030 range from 90,000 to 80,000. AI may account for some of that change. BT hasn’t said how much.

Until it does, anyone counting telco jobs lost to AI should count most of BT’s as lost to fiber, a one-time network upgrade that needs fewer engineers to keep running. Telcos that moved their customers off copper years ago can’t repeat it.

*

Methodology

BT’s “total labour resource” includes employees and sub-contractors, measured as full-time equivalents at fiscal year end (31 March). FY2024-FY2026 figures come from BT’s FY2026 results presentation, and the 130,000 FY2023 baseline from its FY2023 results release. The Telco Workforce Monitor series counts employees only, so it matches BT’s direct labor but is lower than “total labor resource”. The 10,000 roles tied to digitization and AI, and the 15,000, 10,000 and 5,000 figures for fiber build, maintenance and other functions, come from Philip Jansen and BT statements to The Register in May 2023. The Monitor’s BT figures between BT’s reported dates, including 2Q26, are MTN Consulting estimates. The 6.9% median covers the 72 telcos with data in both 2Q22 and 2Q26. BT publishes no headcount by unit, so the shares of its actual cuts due to fiber and to AI are not reported.

About Gen AI Telco

The Gen AI Telco Substack is published by MTN Consulting LLC. Founded in 2017, MTN Consulting is a boutique advisory firm focused on telecom and hyperscale markets. We work with companies, investors, and policymakers on AI transition topics. Clients, soon-to-be clients, and journalists, feel free to contact us with any questions: Chief Analyst Matt Walker.

Share

But this program is aimed at the thousands of workers laid off by Verizon. The company has not publicly aimed any of the $70M at the roughly 97,600 people who still work there. It last put a number on training its own staff in 2023, and it was small: the equivalent of about 1% of capex. Verizon has stopped publishing that number since. By Schulman’s own yardstick, Verizon is leaving out the people it already employs.

Subscribe now

Verizon aimed every dollar at outsiders or people it laid off

Verizon split the pledge in two. It committed $50M in new money to job seekers, early-career professionals, displaced workers, educators and small businesses, with course content from IBM, Google, Microsoft, Anthropic, Coursera and OpenAI. The other $20M is a Reskilling and Career Transition Fund for departing workers, which Verizon created on 20 November 2025, when Schulman told staff it would cut more than 13,000 jobs.

Verizon’s announcement does not mention training current employees. It names no headcount target and no timeframe for the $50M. The program reads as a public-facing campaign: Verizon as the helper for those displaced by AI, but not necessarily for the people still doing the work.

Verizon stopped reporting what it spends training its own staff

Verizon used to publish what it spent on those employees. In its 2023 ESG report, Verizon put employee learning and development at $173.8M that year, including more than $20M of tuition assistance used by about 4,700 employees. That works out to about $1,650 per employee.

Since then Verizon has gone quiet. In its 2024 Responsible Business Update, the most recent, Verizon discloses none of five standard workforce measures, according to a September 2026 MTN Consulting review of the full report. The five are new hires, turnover, training hours, training spend and participation in reskilling programs. Training hours are a standard indicator (GRI 404-1) under the Global Reporting Initiative framework, and many of Verizon’s peers publish them.

Measured against Schulman’s own benchmark, technology spending, Verizon’s 2023 training budget was small. Verizon spent $18.8 billion (B) on capex that year, according to MTN Consulting’s Telco Workforce Monitor. The $173.8M training budget was 0.9% of that. For every dollar Verizon spent training its people, it spent about $108 on networks and equipment.

AT&T, Deutsche Telekom and Vodafone publish how they train staff for AI

AT&T, Verizon’s closest US rival, still reports these numbers. In 2025 it invested over $89M in training, and more than 147,000 employees completed 5.8 million hours of it. It launched an Ask AT&T Academy to teach staff its internal generative AI tool and gave 20,000 employees Microsoft Copilot licenses. AT&T’s level of training investment is even lower than Verizon, as a percent of capex, but at least it publishes the data.

European telcos report AI training for their own staff in the same way. Deutsche Telekom says 30,000 employees outside T-Mobile US acquired AI prompting skills in 2025. Vodafone says 1,947 employees were upskilled toward agentic AI in the year to March 2026, and that it spent an average of €352 per employee on training.

A community skills pledge is not an HR budget, and Verizon almost certainly still trains its staff. But it stopped publishing the figure while its rivals kept going. The only training money Verizon quantifies publicly is directed at outsiders and people on their way out.

An expensive workforce is the one most worth retraining

The case for training existing staff is strongest where each employee costs the most. Verizon’s all-in labor cost per employee ran $155,400 a year over the four quarters to 2Q26, nearly double the $79,900 average across 40 telcos in the Americas tracked in the Monitor. A 1% productivity gain across Verizon’s $15.2B labor bill would be worth about $152M a year, close to its entire disclosed 2023 training budget.

Verizon took on 13,000 Frontier employees and left them out of the pledge

Read more

Yes, it’s true: in the old days of 2G/3G mobile and the PSTN (public switched telephone network), telcos commonly charged for service on a minutes-of-use basis. That was largely abandoned by the early 2010s. Voice calls became effectively free on landlines and mobiles alike, and few people think about their cost now. The meter moved to data, where download caps still ration heavy users. With AI, telcos may come full circle, but this time charge for tokens.

Dropping metered pricing cut the link between how much customers used and how much telcos earned. In China, mobile data traffic grew 17.3% in 2025 while telecom revenue grew 0.7%, according to the Ministry of Industry and Information Technology. Tokens put a meter back on usage.

China Telecom is a standout in the telecom AI world. It is trying to earn money from AI directly instead of just carrying traffic for the companies that do. China Telecom’s core business has stopped growing, as China’s AI providers are growing rapidly. Tokens, the units AI models bill by, give this telco giant a product to sell into that market. The test is whether it can sell them at a profit, because if it can, other telcos have a model worth copying.

China Telecom’s profit fell 14.9% in 1H26 on revenue down 3.9%

In the company’s 1H26 interim results, revenue fell 3.9% to RMB259.0 billion (B) and net profit fell 14.9% to RMB19.59B. The company blamed slowing legacy businesses and a value-added tax (VAT) reclassification, and said core revenue held steady on a comparable basis.

A stalled top line hurts more because the company’s profit margins are razor thin. China Telecom’s margin on earnings before interest and taxes (EBIT) is 6.6%. That is about half the 12.6% average across 49 Asian telcos in MTN Consulting’s Telco Workforce Monitor; see figure, below. Much of China Telecom’s cost base is fixed. Labor takes 25.4% of its operating costs excluding depreciation and amortization, against 21.4% for Asian peers. Headcount has held near 278,000 for five years. With costs that sticky, margin improves mainly through new revenue.

Source: MTN Consulting’s Telco Workforce Monitor, Figure 13, four quarters to 2Q26, US dollars. Solid line: China Telecom. Dashed line: Asia average.

China’s cloud market is growing, but the big cloud providers are growing faster than the telcos. Alibaba’s AI cloud and compute revenue grew 45% in the June quarter, versus just 7.8% growth in 1H26 for China Telecom’s Tianyi Cloud. Alibaba and China Telecom reported the figures in August.

China Telecom is selling AI the way it once sold minutes

To compete for that spending, China Telecom has brought back the meter. It began selling token packages in May, from RMB9.9 a month for 10 million tokens to RMB299.9 for 250 million. Those monthly allowances resemble a mobile data bucket more than a per-minute meter.

The first buyers have come through China Telecom’s own products. Its TeleAgent office assistant has signed up 1.2 million users since July and consumes more than 200 billion tokens a day. China Telecom’s capex budget is supporting this transition. It plans to raise computing capex 26% in 2026 YoY, to 35% of the total, while traditional network spending falls 26%.

The payoff so far is modest. China Telecom groups everything beyond communications services as intelligence revenue. That line grew 7.1% to RMB31.1B in 1H26, 12.8% of core revenue. TeleAgent’s tokens are a sliver of it.

TeleAgent’s token volume would bill under 0.04% of revenue

At China Telecom’s own list prices, TeleAgent’s 200 billion daily tokens would bill RMB46M to RMB194M a year. That is under 0.04% of 2025 revenue.

A skeptic would call that a rounding error, and for now it is one. China’s token market is already large: ByteDance’s Doubao models handled 180 trillion tokens a day in June, Volcano Engine president Tan Dai said. Volume alone still won’t make China Telecom’s version profitable, because token prices keep falling. Model prices have declined roughly tenfold each year, Bain & Company found. A telco reselling other companies’ models may find itself in a race to the bottom.

Beyond raw tokens, China Telecom sells software built on them. It has built 420 industry agents for government, transport and manufacturing customers. An agent is a program that uses AI models to complete a task on its own, rather than answer a single prompt.

Other telcos can copy the pricing more easily than the position

For other telcos, copying the pricing is the easy part. China Mobile and China Unicom have already followed China Telecom into token sales. Copying the position outside China is harder. A European or US telco would sell tokens against Amazon Bedrock, Microsoft Azure AI Foundry and Google Vertex AI. Each already offers enterprises a menu of leading models.

Whoever sells them, tokens reverse a decade of telco pricing. Telcos spent the 2010s teaching customers to stop counting minutes. China Telecom now wants them to count tokens.

This could be a plus. The old calling plans were a chore, but anyone could check a bill against a published rate. AI pricing today offers no such check. As a frequent user of AI Cowork tools, I support any change to bring greater transparency to AI pricing. It would help consumers struggling to understand their bills, and AI providers as well. Ultimately it’s more effective to lock in users based on loyalty and high satisfaction rates, rather than just the fear of having to start work over with another provider.

*

Methodology

Margin, labor and headcount figures come from the MTN Consulting Telco Workforce Monitor, 2Q26 edition, in US dollars over the four quarters to 2Q26. The Asia benchmark covers 49 telcos. Renminbi figures are company disclosures as reported by the named outlets. The TeleAgent billing range is our estimate: 200 billion tokens a day for a year, priced at the lowest and highest per-token rates across China Telecom’s published packages.

About Gen AI Telco

The Gen AI Telco Substack is published by MTN Consulting LLC. Founded in 2017, MTN Consulting is a boutique advisory firm focused on telecom and hyperscale markets. We work with companies, investors, and policymakers on AI transition topics. Clients, soon-to-be clients, and journalists, feel free to contact us with any questions: Chief Analyst Matt Walker.

Share

The two sectors moved in opposite directions for a decade before they crossed. Hyperscaler headcount is up 185% since the first quarter of 2014, from 1.68 million to 4.78 million. Telco headcount is down 16.5% over the same twelve years, from 5.17 million to 4.31 million.

Among hyperscalers, Microsoft discloses the most about its workforce

Telecom business models are relatively homogenous, whereas hyperscalers are all over the place. Some hyperscalers are rooted in ecommerce, some in electronic devices, some in software, some in social media. The sectors have some crossover in technical skills required. For instance, as hyperscalers began investing in subsea cables, they hired talent away from the telecom sector aggressively in order to do so. As AI has exploded, some high-profile telecom execs have migrated; Rakuten’s ex-CTO Tareq Amin, for instance, is now in charge at the Saudi AI giant HUMAIN. But in general they hyperscalers don’t disclose a lot of details on their workforces, relative to telcos.

Microsoft is the only one of Amazon, Microsoft, Alphabet and Meta whose annual filing breaks employees out by function. Of 228,000 employees as of 30 June 2025, Microsoft puts 80,000 (35.1%) in product research and development, 89,000 (39.0%) in operations, covering data centers, product support and consulting, and manufacturing and distribution, 44,000 (19.3%) in sales and marketing, and 15,000 (6.6%) in general and administration (Microsoft Corporation, Form 10-K for the fiscal year ended June 30, 2025, filed July 30, 2025).

Amazon, Alphabet and Meta give only a total headcount in their 10-Ks, and their most recent sustainability reports add no function breakdown either: Amazon at 1.58 million as of 31 December 2025, Alphabet at 190,820 as of 31 December 2025, and Meta at 78,865 as of 31 December 2025. Alphabet’s environmental report does not restate a headcount figure at all, and Meta’s sustainability report discusses supply-chain labor practices rather than its own workforce composition.

Telcos that disclose a split put most employees in sales and customer care, not operations

Vodafone and China Telecom are two telcos that publish a comparable split. Vodafone’s fiscal 2026 annual report puts 61.2% of its 92,988 employees in customer care and administration, 22.3% in selling and distribution, and 16.5% in operations (Vodafone Group Plc, Annual Report FY2026). Two years earlier, Vodafone’s FY2024 annual report showed a similar shape at a larger headcount: operations 16.3%, selling and distribution 23.8%, customer care and administration 59.9% of 96,282 employees. Vodafone’s total headcount fell 3.4% between the two reports, and the customer-care share of what remained rose, not fell.

China Telecom’s 2025 annual report splits 277,911 employees into sales and marketing (37.7%), operations and maintenance (27.5%), management, finance and administration (17.8%), and sci-tech research and product development (17.0%). China Unicom’s 2025 annual report gives shares only: marketing 51.1%, technical 28.9%, management and support 20.6%.

Added together, Vodafone and China Telecom assign 24.7% of their combined 370,899 employees to operations and maintenance. Microsoft alone assigns 35.1% of a much smaller workforce to research and development. The two telcos put the larger share of their disclosed headcount into sales, customer care and administration, the commercial and service side of the business. Microsoft’s largest single category is the most technical one, R&D.

These are individual company disclosures, not industry averages. A breakdown of this kind is rare on both sides. Of the sixteen hyperscalers and telcos checked for this piece, five gave a function-based employee split and two more split by management level instead of by function. Only three of the five, Microsoft, Vodafone and China Telecom, gave shares that reconcile to a full headcount total.

The crossover measures two kinds of organization, not one workforce moving

The disclosed splits point away from a simple story. Vodafone’s own two-point trend puts its shrinking headcount in selling and distribution, not operations: the operations share held roughly flat while the total workforce fell 3.4% between FY2024 and FY2026. Microsoft’s single disclosed snapshot puts 74% of a much smaller headcount in research and development and operations combined. Someone pricing the telco-hyperscaler crossover as network engineers moving into cloud and AI jobs is pricing a headcount total, not the categories the two companies actually report. The telco losses look commercial, and the hyperscaler gains sit in technical work each company already does.

Share

*

Methodology

MTN Consulting tracks employee counts for hyperscalers and telcos inside two separate quarterly research products, one covering telecom company financials and one covering hyperscale company financials, not a single combined tracker. The panel covers 145 telcos and 25 hyperscalers, not all of them still operating independently today. Where one tracked company is absorbed by another, MTN Consulting moves its employees into the acquirer’s line from the deal’s close and holds the absorbed company’s own figure at zero going forward; its historic figures stay in the panel rather than being deleted or carried forward as if it still reported separately. The 468,000-job gap and the twelve-year trend reflect that continuous, absorption-adjusted panel, not a fixed set of companies.

Questions about how to work with MTN Consulting on telecom AI, automation and autonomous-network research? Contact Matt Walker: matt@mtn-c.com | www.mtn-c.com

Subscribe now

Reliance Industries’ Jio is often cited as proof of the conglomerate model. Reliance frequently highlights these synergies on earnings calls. Jio’s operational efficiency reflects clear cross-pollination from its parent group, alongside an aggressive approach to recruiting international talent that sets it apart from legacy carriers.

Jio got funded for its initial launch for very practical reasons: nothing else in Reliance’s portfolio offered a better use for the money. Eventually it became Reliance’s best-performing segment. Rakuten Group is hoping for the same result with Rakuten Mobile. Six years later, though, the logic has produced the opposite result, as Rakuten continues to report losses.

Both Jio and Rakuten are considered greenfield operators, as they built most of their network from scratch. There are also some large telcos that are part of larger corporate groups, including through direct ownership (Comcast), and minority equity stakes (Singtel).

Jio reached positive earnings before interest, tax, depreciation and amortization (EBITDA) within a year of launch

Subscribe now

Reliance Jio Infocomm launched commercial service on 5 September 2016. By the quarter ended 30 September 2017, twelve months later, Jio reported EBITDA of 14.43 billion rupees. EBITDA excludes depreciation, the dominant cost for a mobile network still being built, so a positive figure alone does not show the business covering its full costs. Reliance’s presentation for that quarter also reported positive earnings before interest and tax, the profit measure after depreciation. Three months later, in the quarter ended 31 December 2017, Jio reported its first net profit after tax: 5.04 billion rupees, fifteen months after launch. Jio has stayed profitable since and now runs the highest margin of any Reliance segment: a 51.9% EBITDA margin in fiscal 2026, against 8.3% for Retail and 9.1% for oil-to-chemicals.

Rakuten Mobile is 6.2 years past launch and still lost 33.1 billion yen last quarter

Rakuten Mobile launched full commercial service on 8 April 2020. In the quarter ended 30 June 2026, 74 months later, Rakuten’s Mobile segment still posted a non-GAAP operating loss of 33.1 billion yen. Table 2 shows how that loss has moved since Rakuten Mobile’s first full fiscal year.

The Mobile segment’s full-year non-GAAP operating loss has narrowed every year since fiscal 2022. Table 2 breaks out the figures.

Estimate, our own calculation: the pace of improvement slowed in fiscal 2025. The loss fell 22.5%, versus 38.1% the year before, a smaller improvement in percentage and in absolute yen (47.1 billion yen, against 128.6 billion the year before). Fiscal 2025 was also the first year Mobile reported a full-year EBITDA profit, 12.9 billion yen, since Rakuten entered the mobile business.

Rakuten’s restated year-on-year figures differ from what it first reported by 14 to 23 billion yen a year. This analysis uses the figures as first reported.

Rakuten entered a mature market under different structural conditions

India had poor mobile data coverage and weak carrier loyalty when Jio launched in 2016, the conditions a price disruptor needs. Japan’s three carriers already gave customers good service and strong loyalty when Rakuten launched in 2020. Japan’s regulator had also already forced tariff cuts of roughly 40% and dropped cancellation fees before Rakuten arrived, narrowing the price gap Rakuten needed to undercut.

Rakuten also launched with a thinner spectrum allocation than the incumbents held. It had to build far more base stations than planned to cover the gaps, which raised network costs for years.

Rakuten’s own case for staying in is not a mobile profitability forecast. Its January 2026 strategy update frames continued investment around “ecosystem synergy,” mobile driving use of Rakuten’s e-commerce and financial services. Rakuten has not published a yen figure for that synergy in any release checked for this piece.

Two other Rakuten businesses are funding Mobile’s continued loss

Rakuten Group posted its first quarterly net profit in six years in the quarter ended 30 June 2026: 7.7 billion yen, an improvement of 58.7 billion yen year on year. Internet Services, mainly e-commerce, contributed 23.1 billion yen in non-GAAP operating income, up 68.6% year on year, and FinTech contributed 69.2 billion yen, up 60.1%. Mobile’s 33.1 billion yen loss was the only segment in the red. On the same earnings call, Rakuten’s finance chief described Mobile’s funding as self-funding, no longer requiring outside financing, according to a transcript of the call published by Investing.com. Rakuten has not disclosed a target date for Mobile to reach a full-year operating profit.

Singtel monetized a minority stake to fund core infrastructure

Singtel and Comcast apply the same logic to telecom assets they do not run outright. Singtel holds a different kind of telecom exposure: minority equity stakes in independently run, separately listed sister telcos rather than a wholly owned network. Its effective stake in India’s Bharti Airtel fell from 31.4% in 2022 to 26.9% as of 30 June 2026, a cut of 4.5 percentage points, funded by a series of stake sales that Singtel has used to pay for its own 5G and data-center spending. Bharti Airtel is not underperforming. Singtel is redirecting capital away from a minority telecom holding toward its own core network spending, a different mechanism from Rakuten funding Mobile’s losses, but the same underlying behavior: a diversified owner moving capital toward whichever part of the portfolio it judges will earn the best return next, not toward telecom by default.

Comcast shed media assets to isolate its broadband and wireless core

Comcast took a structural approach: on January 2, 2026, it completed the spin-off of Versant Media Group ($7\text{ billion}$ in FY2024 revenue, declining at 6.9% CAGR since FY2018). Comcast retained its core broadband and wireless assets (Xfinity, Comcast Business, Sky) alongside NBCUniversal streaming and parks. Management did not exit telecom; it carved out declining media assets to protect core connectivity margins.

The conglomerate is managing telecom as a capital allocation, not a strategic commitment

Four diversified owners, four different-looking decisions, one common mechanism. Reliance kept funding Jio because it paid off inside sixteen months and has kept paying off since. Rakuten keeps funding Mobile because two other segments can currently cover the loss, not because Mobile has hit Jio’s timeline. Singtel is pulling capital out of a telecom stake that is performing fine, because its own network spending currently ranks higher. Comcast protected its telecom business by removing a worse-performing business from next to it, not by protecting telecom for its own sake.

None of these four decisions was made because telecom is strategically special to a diversified parent. Each was made the way a portfolio manager reallocates between assets: fund the one with the fastest credible path to return, trim the one competing with a higher-priority use of capital, and cut the one dragging on the rest, whether or not it happens to be the telecom line.

Two of these bets added a new network to a market that already had enough capacity. Jio and Rakuten Mobile both built from scratch and competed on price, forcing incumbents to answer with margin cuts of their own. Singtel and Comcast never added capacity. They moved capital around telecom assets that already existed. That distinction is about to be tested again.

SpaceX is choosing whether to become a fourth U.S. carrier or buy one of the three

SpaceX already has spectrum and a paying mobile business. It bought $17 billion of spectrum from EchoStar in September 2025 to expand Starlink’s Direct to Cell service. Semafor reported on 29 July 2026 that SpaceX is now seeking more spectrum suited to dense urban areas, by acquisition or at auction. Starlink’s mobile business is projected to reach $15 billion in revenue this year.

Musk said in September 2025 that buying a carrier outright was “not out of the question” (Semafor). Wolfe Research analyst Peter Supino has gone further, arguing publicly that SpaceX should buy an existing carrier, preferably T-Mobile, rather than build an independent fourth network (Fierce Network). His reasoning: a new price-competing entrant would cut industry profitability more than owning an existing carrier would, for SpaceX included.

I now weight SpaceX’s odds of getting a carrier deal past this FCC up, not down, on politics. Musk is not Paramount, buying access for the first time with a one-off settlement. He was the single largest donor of the 2024 election, spending at least $277 million to help elect Trump, and ran DOGE, the Department of Government Efficiency, from its creation on 20 January 2025 until he stepped down that May. A public feud followed in mid-2025, over Trump’s tax-and-spending bill and Musk’s comments about Jeffrey Epstein. Both men have been rebuilding the relationship since: Musk joined Trump’s Beijing trip in May 2026 and had pledged at least $100 million to Republican midterm campaigns by August.

SpaceX’s strategic choice—building a fourth standalone network versus acquiring an existing operator—will determine the market impact. Building a fourth network would mirror Jio’s entry in India, diluting market margins across all operators. Buying an incumbent maintains industry capacity while transferring ownership.

Investors evaluating telecom entry must price Jio’s timeline, not its mythology

Reliance is repeatedly cited as proof that a conglomerate can fund a loss-making mobile network into a profit center. The number usually left out is how fast that happened: twelve months to positive EBITDA, fifteen months to net profit. Rakuten Mobile, funded on the same premise by the same kind of diversified parent, is 74 months past launch and has not yet posted a full-year operating profit. Its loss is narrowing, but more slowly than it was two years ago, and Rakuten has not said when, or whether, that changes.

None of these four cases shows an ownership structure that creates telecom growth. Building a new network grew revenue by cutting competitor margins, not by expanding the market. If SpaceX buys an existing carrier, it replaces an owner without adding revenue. If it builds a fourth network, it adds a competitor, and incumbents take the same margin hit Reliance’s rivals took.

Subscribe now

About Gen AI Telco

The Gen AI Telco Substack is published by MTN Consulting LLC. Founded in 2017, MTN Consulting is a boutique advisory firm focused on telecom and hyperscale markets. We work with companies, investors, and policymakers on AI transition topics. Clients, soon-to-be clients, and journalists, feel free to contact us with any questions: Chief Analyst Matt Walker.

Share

e& is an extreme case, but not an outlier. Across 29 operators with usable Apollo data, retail and sales roles make up 32.1% of open postings (2,005 of 6,254). AI and machine learning roles make up 3.2% (201 of 6,254), roughly a tenth as many. The unweighted average across all 29, each counted once regardless of size, is 28.3% retail and sales against 1.9% AI and machine learning. This isn’t Verizon, AT&T and Orange (40% of the pool between them) skewing the result. Twelve of the 29 operators posted zero AI or machine learning roles at all.

Titles undercount AI hiring. The real spread is basic AI literacy, not specialist roles.

To conduct our analysis, the first step was to sort every posting into one category from its title. We gave AI/ML top priority, so a posting mentioning both “AI” and “sales” counts as AI. That makes 3.2% generous, not conservative: folding the entire software/data/cloud category in as an upper bound still only reaches 9.3% against 32.1% for retail and sales.

Focusing narrowly on so-called “AI jobs” is misleading, though. The OECD’s June 2026 policy brief on AI and skills puts a number on how much of the workforce actually needs deep AI skills, not just AI awareness. Per an older OECD report, fewer than 1% of workers need advanced AI-specific skills such as programming or model development (Green and Lamby, 2023), the figure the brief itself still relies on. AI/ML job titles are that 1%. The OECD brief reports that what’s actually spreading through ordinary jobs is digital skills and the ability to use, analyze and interpret data, not a title change. A posting for a store manager or an account executive that lists comfort with AI tools in its qualifications never shows up in the AI/ML column above. The 3.2% figure measures how many jobs are all about AI. It doesn’t measure how many now need some AI-adjacent skill to do at all.

Source: Apollo.io Organizations Job Postings API, pulled 8-15 September 2026, one snapshot per operator.

Cox Communications’ 92.9% retail-and-sales share is the highest in the panel, on the smallest sample (n=14). Cox is privately held, sits outside MTN Consulting’s public-company workforce panel, and was acquired by Charter Communications in August 2026. Its independent hiring has a shrinking shelf life regardless of the mix. One interpretation: when a company is in the process of being acquired, they still need to hire front-line sales staff, but often hold off on management hiring. AT&T (70.6%) and Bell Canada (62.2%) are the two next-highest, both large consumer operators with extensive owned retail. Singtel (6.4%) and KT (6.2%) post the highest AI/ML shares in the panel. KT’s sample includes one title using AX, the company’s own internal shorthand for AI Transformation. We caught it and excluded it anyway: AX is KT’s internal shorthand for AI Transformation, not a term an outside reader would recognize as AI, so counting it would have inflated KT’s AI share on a technicality.

AT&T cut headcount 17% since 2022. Verizon cut 23%, but added some back by buying Frontier.

Both of the two largest US carriers have cut deeply since 2022. Per AT&T’s Corporate Responsibility KPI page, global headcount was 160,700 in 2022, 149,900 in 2023, 140,990 in 2024 and 133,030 in 2025, a decline of 27,670 employees, or 17.2%, over three years. That baseline already excludes WarnerMedia, Vrio, Xandr and Playdemic: AT&T’s 2022 annual report classifies all four as discontinued operations, divested before the 160,700 figure was tallied, so the decline isn’t a spinoff artifact.

Verizon’s decline is just as steep. Per MTN Consulting’s Telco Workforce Tracker, 2Q26 edition, confirmed against Verizon’s own 10-K and earnings releases, total employees fell from 117,100 in 2022 to 89,900 at the end of 2025, a 23.2% drop. Verizon’s fourth-quarter 2025 release ties the cut to $2.156 billion in full-year severance charges. Headcount then jumped to 99,600 by the first quarter of 2026, not from a hiring rebound but from the Frontier Communications acquisition, which closed 20 January 2026 and added Frontier’s roughly 13,000 employees. By the second quarter, the combined company was back down to 97,600.

e& Group’s single open role is a good illustration of a broader trend. Across the panel, telecom’s job boards are still actively looking for store staff and account managers. These roles are far more prominent than the AI engineers the industry keeps talking about in its earnings calls. AT&T and Verizon are proof that the industry can shed a fifth of its workforce without any of it showing up as an “AI jobs” story. Both companies point to efficiency and technology in their earnings calls. Neither has posted the AI hiring to match the cuts.

*

Methodology note: Job postings are sorted into one category per title, AI/ML first in priority order. LG Uplus, NTT Docomo Business and PLDT returned zero postings and are excluded from every figure above; Bharti Airtel wasn’t pulled because Apollo splits it across a dozen-plus entities with no clean parent. All figures are Reported: the Apollo pull (8-15 September 2026), e&’s deal terms (its own board announcement), and AT&T’s and Verizon’s headcount (company filings and MTN Consulting’s Telco Workforce Tracker).

Subscribe now

Note to readers:

For more information about MTN Consulting’s telecom workforce research, contact Matt Walker: matt@mtn-c.com. MTN Consulting’s Telco Workforce Tracker carries the AT&T and Verizon headcount series behind this piece, along with over 140 other telcos across the globe. Full details in the 2Q26 edition.

*

THE UNMANNED NETWORK | Covering 5 August to 13 September 2026

Bottom Line Up Front

Covering August 5 to September 13, 2026, this issue tracks telecom operators reporting AI as a distinct revenue line rather than an internal cost-cutting tool. It also examines the industry’s first attempts to benchmark “AI-native” operators.

China’s three state-owned operators gave the clearest revenue numbers. China Telecom, China Mobile and China Unicom reported first-half 2026 (1H26) AI-linked computing revenue growth of 13% to 95%, built on token-subscription platforms running on their own state-controlled large language models (LLMs). This is something other telcos could copy. Each of the three runs a marketplace where an enterprise buys access to models by the token - the unit of text a model reads or writes - in the same way it once bought minutes or megabytes. China Telecom’s hosts 142 models and more than 420 industry applications. China Unicom calls its version a “token supermarket.” That turns AI compute from a capital project into a metered product with a per-unit price.

Outside China, SK Telecom reported Q2 2026 AI data center (AIDC) revenue of KRW 136.2 billion ($90 million, +92.5% YoY). In the UAE, e& bundled agentic AI directly into business mobile and broadband tariffs. Modernization cost-cuts continued in parallel: T-Mobile US eliminated over 4,500 roles in 2026 as parent Deutsche Telekom accelerated AI deployments.

Benchmark methodologies also debuted. TelecomTV’s initial AI-Native Telco Index scored 56 operators, placing nine in its top tier: AT&T, China Mobile, Deutsche Telekom, KDDI, NTT/Docomo, Rakuten Mobile, SoftBank, Telefónica, and Verizon. Crucially, of the 20 operators prioritized for AI revenue generation, 15 are in Asia and the Middle East, and 11 are Tier 2 operators ($2B–$10B revenue). Western incumbents lag in top-line commercialization.

Meanwhile, vendors announced over $1 billion in AI-tagged network contracts across Indonesia and Costa Rica. Vendors continue to link AI to new features in press releases announcing contract wins, and we expect this to continue. Whether the AI link is fact or fiction is another question.

Subscribe now

Signal vs. Noise

Below is a summary of recent developments in telecom AI, automation and autonomous networks, our view on why they matter, and their implications for operators, vendors and investors.

China’s three state telcos turned AI compute into a reported revenue line, up 13% to 95%

21-Aug-26

China Mobile, China Telecom and China Unicom convert token platforms into a reported revenue line

China’s three state-owned operators now report a working method for billing AI compute the way they once billed mobile data.

• China Telecom’s Xingchen TokenHub hosts 142 large language models (LLMs) and more than 420 industry-specific applications. The platform helped drive a 95% jump in intelligent computing revenue and a 7.1% rise in broader intelligent-business revenue to RMB 31.1 billion ($4.6 billion), based on the company’s 1H26 interim report.

• China Unicom’s UniAI platform connects more than 200 LLMs through a token-subscription model it calls a “token supermarket.” The platform contributed to a 13% rise in computing-power revenue to RMB 41.9 billion, a figure confirmed in the company’s own 18 August 2026 interim results presentation.

• China Mobile’s “token factory” strategy and MoMA platform pushed computing services revenue up 14% to RMB 52.9 billion, with intelligent computing services revenue growing 130% to RMB 5.3 billion. Both figures are confirmed in the company’s own 13 August 2026 interim results.

The models behind these platforms differ by operator. A comparison of China Mobile’s Jiutian, China Telecom’s Xingchen, China Unicom’s Yuanjing and China Electronics’ Guanlan found each model built toward a different enterprise vertical, with no single operator’s LLM dominant across all four (Chinese-language source, summarized).

Why it matters: The most important part of this story is that these three operators are doing it at all. No other national operator group has published comparable AI-token revenue figures. If the token-subscription model holds up as these platforms scale, it is a template other operators elsewhere could copy.

These are the first concrete revenue numbers behind a strategy Chinese operators have discussed since early 2026: selling AI compute like a subscription product rather than treating it as a sunk infrastructure cost. The percentage gains vary by more than seven times across the three operators, which suggests the token-subscription model is still maturing rather than proven at uniform scale. None of the three operators discloses token revenue as its own line item. The figures reported here are company-disclosed segment totals for intelligent computing or computing power, not audited, filed breakouts of AI-token sales specifically.

Sources: https://www.chinatelecom-h.com/en/ir/report/interim2026.pdf | https://www.chinamobileltd.com/en/media/press/p260813.pdf | https://www.chinaunicom.com.hk/en/ir/presentations/pre260818.pdf

Three telcos now disclose AI revenue, and the largest of them is C$700 million

10-Sep-26

Bell Canada, SK Telecom and Indosat put figures on the table, while T-Mobile cut 4,500 jobs in the same window

Three more operators have now put a number on AI revenue.

TelecomTV’s AI-Native Telco Index reports that Bell Canada booked around C$700 million (US$507 million) of AI-powered revenue in 2025, up 60% year on year, against a target of about C$2 billion a year by 2028. BCE’s own results release for 2025 gives a growth rate rather than a full-year figure, and the C$2 billion target is confirmed on its 1Q26 earnings call. SK Telecom’s own 2Q26 results release reports AIDC revenue of KRW 136.2 billion ($90 million), up 92.5%, plus AI business and consumer services revenue of KRW 61.3 billion ($40 million), up 24.5%. Indosat Ooredoo Hutchison reports its AI cloud income as a separate line, at $28 million in 2025 with guidance of around $60 million for 2026, a figure its management frames as the top of a range. Singtel described its own AI progress in a public update without attaching a figure.

In our separate story on TelecomTV’s AI Native Index, we note that 20 of 56 telcos are identified as aiming AI primarily at revenue, 28 as balancing revenue and efficiency, and six as primarily chasing efficiency. Its efficiency examples are KPN, Proximus and Spark New Zealand. KPN is targeting about EUR 100 million a year of operating-cost savings by 2030. Proximus runs AI inside a EUR 180 million efficiency program, and Spark is tracking a reduction in customer loss of up to 30%. In the same window T-Mobile US cut more than 4,500 jobs in 2026 as parent Deutsche Telekom pushes AI-driven modernization.

Why it matters: This is the same trend the newsletter’s 5 August issue described, but it now has figures attached. Bell Canada’s C$700 million is the largest AI revenue any telco outside China has disclosed, and it is a low single-digit share of Bell’s total revenue. SK Telecom’s AIDC line is $90 million for the quarter. Indosat’s is $28 million for the year. Set those against the capital going in, including Indosat’s own $800 million Zankore venture and SK Telecom’s multi-gigawatt buildout, and the revenue is one to two orders of magnitude below the spend. The growth rates are real and steep, and the 2028 targets are the thing to hold operators to. Treat the disclosure itself as the signal for now: three operators have decided this line is big enough to report, which is what the previous four quarters lacked.

Sources: https://news.sktelecom.com/en/3218 | https://www.bce.ca/news-and-media/releases/show/bce-reports-2025-q4-and-full-year-results-announces-2026-financial-targets | https://www.telecomtv.com/content/ai/new-report-highlights-telco-focus-on-ai-native-revenue-opportunities-56218/ | https://www.sdxcentral.com/news/t-mobile-us-cut-more-than-4500-jobs-in-2026-as-deutsche-telekom-drives-modernization/ | https://www.telecomtv.com/content/ai/singtel-lays-out-its-ai-progress-ambitions-56141/

A 56-operator index names nine AI-native leaders

07-Sep-26

Two efforts this cycle tried to answer the same question: how do you measure whether an operator is “AI-native,” and who qualifies? TelecomTV assessed 56 operators against published evidence, with a 30 June 2026 cut-off. It published the result as its AI-Native Telco Index on 7 September, then unveiled the 276-page report at its own AI-Native Telco Forum in Düsseldorf over the following two days. It named nine operators as its top archetype, AI Vanguard: AT&T, China Mobile, Deutsche Telekom, KDDI, NTT and NTT Docomo, Rakuten Mobile, SoftBank, Telefonica and Verizon. Five more are Infrastructure Architects and 15 are Strategic Accelerators, across seven archetypes in all, with a single operator left at Transformation Pending.

Separately, FutureNet World asked whether Asia-Pacific operators are ready for fully autonomous networks, a narrower and more concrete question than the global index attempts. That effort found readiness uneven across the region without naming which operators lead or lag.

Turk Telekom did not wait for an index to catch up. It introduced a no-code AI agent platform aimed at level 4-5 network autonomy, the industry’s own top tier for self-managing networks, overseen by the TM Forum. The company positioned the platform partly as reducing dependence on foreign AI technology (Turkish-language source, summarized). While this company was not on TelecomTV’s leader list, it makes a good case for being ranked among the globe’s telco AI leaders.

Why it matters: The new AI native telco index is a meaningful attempt to standardize a term the industry does not yet agree on. The named list of leaders is the useful part: nine operators, heavily weighted to the US, Japan and Germany, with one Chinese operator and none from India, Africa or Latin America. Read that as a map of where published AI evidence exists, not where AI capability exists. Note that TelecomTV runs the index inside its AI-Native Telco Accelerator program, launched in June 2026 with Axiata, Deutsche Telekom, Orange, NTT Docomo, Rakuten Mobile, the GSMA, Wind River and Totogi as founding participants. Three of those founding participants, Deutsche Telekom, NTT Docomo and Rakuten Mobile, are among the nine AI Vanguard.

Sources: https://www.telecomtv.com/content/ai/new-report-identifies-the-ai-native-telco-frontrunners-56184/ | https://www.telecomtv.com/content/ai/ai-native-telcos-prep-major-updates-insights-for-industry-forum-56153/ | https://www.telecomtv.com/content/ai/leading-csps-technology-innovators-and-industry-bodies-back-launch-of-industry-wide-ai-native-telco-accelerator-55746/ | https://www.telecomtv.com/content/ai/anta-builds-momentum-with-seven-new-partners-56105/ | https://futurenetworld.net/are-asia-pacific-telecom-operators-ready-for-fully-autonomous-networks/2026/09/ | https://www.dunya.com/sektorler/bilim-ve-teknoloji/turk-telekomdan-sektore-yon-veren-oncu-hamle-yapay-zeka-ile-tam-otonom-aglar-donemi-haberi-837920

e& puts agentic AI inside a paid business plan, the cycle’s only AI feature with a price attached

18-Aug-26

The UAE operator bundles agentic AI into Business Mobile and Business Pro Internet, paired with domestic sovereign compute

UAE operator e& (Etisalat) began bundling agentic AI directly into its Business Mobile and Business Pro Internet plans. Agentic AI here means automated task execution, workflow management and information retrieval carried out by software agents acting on the customer’s behalf without a human approving each step. The bundle is paired with UAE-based sovereign AI compute, meaning the processing stays on infrastructure inside the country and under its jurisdiction, which matters to government and regulated enterprise buyers. e& disclosed no price premium, no attach rate and no revenue figure for the bundle.

Why it matters: This is the only item this cycle where AI is a billed product attribute rather than an internal efficiency project or a capital commitment. That distinction is the one investors should hold onto: an operator that charges for AI has a demand signal, however small, while an operator that builds AI capacity has a cost and a hypothesis. It’s too early to know what the revenue impact is, though. The test to watch is whether e& or any peer discloses an attach rate or an average revenue per user effect at its next results.

Source: https://www.telecomstechnews.com/news/e-uae-embeds-agentic-ai-into-mobile-and-broadband-products/

Ericsson and Amdocs both win AI-linked contracts, at opposite ends of the network

Aug 2026

A $220 million radio network build in Costa Rica and a billing modernization in Chile, sold on the same automation pitch

Costa Rica’s state utility ICE (Instituto Costarricense de Electricidad) awarded Ericsson a $220 million contract to deploy a standalone 5G network using AI-powered EIAP rApps, automated software components that manage radio access network (RAN) functions. Movistar Chile agreed with Amdocs to modernize its billing and operating systems (BSS/OSS) using automation and AI, disclosing no contract value.

Why it matters: Automation and AI now sell across the whole telco network, not one corner of it. These two contracts sit at opposite ends: Ericsson in the radio access network, Amdocs in billing and operations support. The vendors benefiting are spread just as widely, which is why this newsletter tracks the deals rather than any single supplier. For operators, the practical read is that AI capability is becoming a standard feature of the products they were going to buy anyway, in the RAN and in BSS/OSS alike.

Sources: https://www.ericsson.com/en/press-releases/latin-america/2026/ice-selects-ericsson-to-deploy-costa-ricas-5g-network | https://www.tabulado.net/movistar-chile-acuerda-con-amdocs-modernizar-sus-sistemas-bss-y-oss-con-automatizacion-e-ia/

Indosat and SK Telecom are building AI compute to sell, at 1 GW and a targeted 15 GW

06-Aug-26 (Indosat); Korea buildout dated separately below

Two operators are funding merchant AI capacity rather than buying it, on very different scales

Indonesia’s Indosat Ooredoo Hutchison launched Zankore, a 1 gigawatt (GW) AI infrastructure venture backed by $800 million from parent Ooredoo, aimed at selling AI compute capacity rather than consuming it internally.

South Korea shows what the same strategy looks like several years further along. SK Telecom is targeting 15 GW of AI data center capacity in Korea by 2035, with 5 GW operational by 2029. It has put the buildout into a wholly owned subsidiary, SK Hyper, with a stated KRW 750 billion commitment through 2030. Its Ulsan site is approaching 900 megawatts against an initial 100 megawatts, at a cluster cost SK Group chairman Chey Tae-won put at roughly KRW 7 trillion ($5.2 billion) on 13 September 2026. SK Telecom is an Nvidia Cloud Partner and sells GPU capacity as a service through its Haein cluster. It has not named a third-party compute customer, so its 2Q26 AIDC revenue of KRW 136.2 billion is the only published measure of what the business is currently selling.

Why it matters: Both operators are making the same bet: that enterprises in their home markets will buy AI compute from a domestic telco for sovereignty, latency or procurement reasons rather than from a US- or China-based hyperscaler. SK Telecom’s scale makes it the test case. Its 15 GW by 2035 is a commitment an investor can measure against reality. For vendors, the relevant point is that the buyer for this equipment is now a telco not a hyperscaler. Telcos have different procurement cycles and different regulatory requirements than hyperscalers. Can a telco move at hyperscaler speed? Indosat’s Zankore, at 1 GW and $800 million, is the smaller and more testable version of the same question.

Sources: https://www.fierce-network.com/cloud/indosat-launches-1-gw-ai-infra-play-800m-ooredoo-backing | https://news.sktelecom.com/en/3155 | https://nvidianews.nvidia.com/news/sk-telecom-ai-infrastructure | https://www.koreatimes.co.kr/business/tech-science/20260913/sk-chairman-says-ulsan-ai-data-center-expansion-nearing-900-megawatts

Two bodies moved to standardize the infrastructure under agentic AI, not the AI itself

28-Aug-26

A network standard and an identity framework target the layer under the AI

Two standards-adjacent moves this cycle target the network infrastructure that agentic AI runs on.

The International Telecommunication Union’s telecommunication standardization sector (ITU-T) approved Y.3335, a framework for low-latency, energy-efficient networks intended to support AI-era traffic. The US National Institute of Standards and Technology (NIST) and a group of European telcos converged on digital identity as a required foundation for agentic AI.

Why it matters: Neither of these two moves is itself an AI product. They are the standards and infrastructure layer a vendor or operator needs before it can sell an AI-native service with confidence it will interoperate or authenticate correctly. The identity convergence matters most for operators already selling agentic AI features to consumers or businesses, e& among them: an agent acting on a customer’s behalf without a verifiable identity chain is an impersonation and liability risk. Y.3335 matters most for network equipment vendors and operators building AI-era transport capacity: without a shared benchmark, a vendor’s claim to build low-latency, energy-efficient AI networking has nothing to be measured against.

Sources: https://www.telecomtv.com/content/access-evolution/itu-t-standardises-framework-of-low-latency-and-energy-efficient-networks-to-support-the-ai-era-56130/ | https://www.biometricupdate.com/202608/nist-european-telcos-converge-on-identity-as-foundation-for-agentic-ai

Equinix links automated interconnection to 8 of the top 10 AI model providers

02-Sep-26

Equinix lets a customer describe what it wants in plain language, then automates the order and its execution; its one named customer is an automaker’s IT unit, not a telco, but that may change

Equinix recently launched Fabric One, a service that lets a customer describe a connectivity need through a portal, an API, or a natural-language prompt, and have the system automate the routing, encryption, resiliency and failover that connection requires. This is a big deal: if it works, a customer can simply describe what it wants in an LLM prompt, and Equinix will implement automatically. Equinix cites scale behind the service: more than 10,500 interconnected businesses, roughly 3,000 cloud and IT providers, 8 of the top 10 AI model providers and 9 of the top 10 neoclouds active on its platform. Named integration partners are AWS and Google Cloud, built on open OpenAPI 3.0 interconnect specifications. The service is pre-release: a beta is planned later in 2026, with general availability in 2027.

Equinix’s only named customer for Fabric One is Hyundai AutoEver America, the automaker’s IT services unit, not a network operator. However, across its operations Equinix has many telco customers, colocating equipment and buying interconnection to link their own points of presence, and mobile edge-compute sites, to hyperscaler and AI-provider on-ramps. Automating that provisioning, rather than having network engineers configure each link by hand, is the kind of use a telecom operator could plausibly adopt for its own wholesale and enterprise interconnection products.

Why it matters: This is genuinely new automation, not a rebrand: intent-driven, natural-language provisioning of multi-cloud interconnection did not exist in Equinix’s portfolio before this announcement. The telco use case above is MTN Consulting’s own example, but it seems plausible to us. Even if not, Equinix has proven the technology can serve one enterprise IT buyer, and should be monitored.

Source: https://newsroom.equinix.com/2026-09-02-Equinix-Unveils-Equinix-Fabric-One,-Redefining-How-Enterprises-Connect-Across-AI,-Cloud-and-Networking-Infrastructure

Comviva is winning awards and ranking on par with much bigger Amdocs in customer reviews

Jul-26 (Comviva’s own newsletter; underlying claims span Jan-Jul 2026)

A Tech Mahindra unit collects six analyst and awards recognitions in a quarter; independent peer-review data show it rating close to Amdocs, a much larger rival

Comviva is a Tech Mahindra unit selling billing and support systems (BSS), digital engagement and payments platforms to more than 130 telcos across 95-plus countries. Its July 2026 newsletter listed six analyst and awards recognitions in 2Q26, all tied to AI or autonomous operations. Juniper Research named it an Established Leader in digital wallets. Forrester included it among eight vendors worldwide on AI agents in customer relationship management (CRM).

Chief Executive Rajesh Chandiramani said in an interview reported around April 2026 that AI is pushing telecom toward autonomous, zero-touch operations. He set a target to reach roughly INR 4,000 crore in revenue, an estimated $480 million, within three years. He described that as doubling. Comviva’s most recent publicly filed financials, in Tech Mahindra’s FY2023-24 subsidiary annual report, show consolidated total income of INR 13,094 million, roughly $157 million. The target is more than triple that disclosed base, not double it, unless revenue has grown substantially since the last filed figure.

MTN Consulting does not formally rate telecom AI vendors, but the independent data is favorable. Gartner Peer Insights community ratings, aggregated by RFP.wiki, put Comviva at 3.7 out of 5 on 75 reviews for AI in telecom customer and business operations. Amdocs, far larger and better known, rates 3.8 out of 5 on 84 reviews in the same market. Comviva’s last disclosed revenue of $157 million in fiscal 2023-24 is less than 4% of Amdocs’ $4.53 billion in fiscal 2025.

Why it matters: There are at least three notable vendors in the AI-in-BSS space for telcos: Comviva, Amdocs and Netcracker. Comviva’s near-parity with Amdocs on customer review scores, despite being a fraction of its revenue and scale, suggests the technical gap between the two is narrower than their size difference implies. An operator shortlisting vendors should ask each for a named reference deployment with results. Comviva’s smaller BSS competitors do not need Comviva’s scale to compete for the same kind of visibility. An active presence in industry awards and analyst reviews is open to a small vendor with a strong submission, as much as to a large one.

Sources: https://www.comviva.com/resources/newsletter/ | https://completeaitraining.com/news/comviva-targets-revenue-doubling-in-three-years-with-ai/ | https://www.rfp.wiki/artificial-intelligence/ai-csp-customer-business-operations/comviva/amdocs | https://insights.techmahindra.com/investors/tml-subsidiary-annual-report-volume-2-2023-24.pdf

Subscribe now

In other news

This cycle (5 Aug-13 Sep 2026)

• Rakuten Mobile integrated Anthropic’s Claude into its Rakuten AI for Business platform, one of the first named large-model deployments inside a telco’s own enterprise product. (6 August 2026). Read more

• KDDI and AVITA trialed a Gemini-powered humanoid robot with Osaka Metro, one of the few physical-robot deployments tracked this cycle. (28 August 2026). Read more

• Zain Kuwait began network testing of real-time AI call translation using a Huawei smartphone, though the announcement discloses no accuracy or latency metrics. (7 August 2026). Read more

• Nokia laid out an AI strategy for Africa’s telecom market as the region enters what the company calls its next growth phase. (1 September 2026). Read more

• Slovakia’s CETIN Networks selected RADCOM for AI-driven network assurance across its mobile network, a deal RADCOM disclosed by customer count rather than contract value. (12 August 2026). Read more

• stc detailed AI partnerships with AMD, Intel, Cisco and AWS spanning data-center capacity, inference compute, edge AI and enterprise adoption, plus tourism and sports use cases. (4 September 2026). Read more

• Indonesia’s RAIA Grid launched to build what its backers call a digital superhighway, a data-transport and compute infrastructure initiative distinct from Indosat’s Zankore venture covered above. (26 August 2026). Read more

Newly identified (other 2026 dates, outside this issue’s review window)

• Amdocs introduced aOS, an agentic operating system for telecom operations. No operator customer or pilot was named. (3 February 2026). Read more

• Alerify and Zadara launched a multi-tenant, NVIDIA-powered sovereign AI cloud for businesses in central Pennsylvania, an enterprise edge deal rather than a telecom operator deployment. (27 January 2026). Read more

• Samsung and Nvidia announced further collaboration on AI-native, software-driven radio access network architecture, naming no operator running the technology commercially. (3 March 2026). Read more

• Ciena unveiled AI-era optical hardware and agentic assurance features, claiming up to 75% lower power and 85% less space for its hyper-rail photonics. No operator customer was named. (10 March 2026). Read more

• Rakuten Mobile and Rakuten Symphony completed a nationwide RAN Intelligent Controller (RIC) deployment with third-party rApp integration, an actual production automation deployment on Rakuten’s own network. (26 February 2026). Read more

• Cisco launched Cloud Control, a unified platform for human and AI-agent network management. CEO Chuck Robbins said network capacity needs could triple within three years. No operator customer was named. (2 June 2026). Read more

• Red Hat published vendor commentary on the TM Forum’s autonomous-network maturity model, describing most providers as still at level 2 of 5. No operator was named. (4 June 2026). Read more

• Marvell announced an AI memory and silicon portfolio for agentic AI inference, claiming 2-3x token throughput gains. It names ‘leading hyperscalers’ generally, not a specific customer. (4 August 2026). Read more

Questions about how to work with MTN Consulting on telecom AI, automation and autonomous-network research? Contact Matt Walker: matt@mtn-c.com | www.mtn-c.com

Share