Almost all telecom earnings calls this year mention AI, yet few mention what the numbers actually show.
Network API revenue is estimated at between $550 million and $2.7 billion globally in 2026, stemming from identity checks, location pings, and quality guarantees which operators sell to developers [9,10]. Global telecom service revenue is close to $1 trillion, and forecasts for direct network API revenue by 2029–2030 disagree sharply on scale relative to that pool: Juniper Research, IDC, Analysys Mason, and Omdia converge on a narrower estimate of roughly $7–9 billion, representing under 1% of total telecom revenue, while McKinsey's own estimate of $10–30 billion in direct API revenue implies closer to 1–3% [6,7]. Even at the high end, network APIs remain a small fraction of the industry's overall revenue base.
A second, larger opportunity is unfolding quietly inside the same companies: monetizing the physical assets which telecom already controls, including power capacity, fiber routes, central office real estate, and sovereign hosting credentials that hyperscalers cannot easily replicate [8].
Look at where capital and revenue targets are landing. Bell Canada raised its AI revenue target to roughly $1.45 billion USD by 2028 and grew AI powered solutions revenue 113% year over year, driven largely by data centers built on its own land and power contracts [1]. TELUS opened a sovereign AI factory in Quebec with clients like Accenture and OpenText, built on the same logic [2]. Further, SK Telecom is filling data centers with GPU as a service demand rather than API traffic [3]. These successes happened even though the industry still hasn’t decided whether Ericsson’s Aduna or Nokia’s Network as Code should serve as the main platform for aggregating telecom network APIs [4,5]. Operators with power and fiber have simply moved past that debate.
The same pattern shows up inside individual enterprise deals and past the corporate balance sheet level. Ericsson's private 5G deployment for Jaguar Land Rover runs on dedicated spectrum and on-site infrastructure rather than a network API [13]. Verizon's private network contract across the Thames Freeport in the UK follows the same model, selling dedicated capacity into a logistics operation rather than a developer platform [14]. In both cases, the enterprise is paying for guaranteed physical performance, and neither deal shows up in anyone's network API revenue line.
The implication for enterprise buyers and infrastructure vendors is direct. When evaluating a telecom partner for AI workloads, the API roadmap matters less than three other questions: What are the details of their power contracts? What are their available fiber routes? Do local regulators trust them to host sovereign workloads? A roadmap slide about network APIs says little about whether a vendor can deliver reliable power at scale or pass sovereign hosting review, and those assets predict who wins this decade far better than developer documentation does.
For operators, the lesson is uncomfortable. Years of investment went into network exposure platforms that generate, by Ericsson's own admission, tiny revenue [4]. The bigger prize was sitting in already owned real estate and power contracts the whole time.
Network APIs still have a real, if narrow, role: fraud prevention APIs already generate repeatable revenue for banks and payment providers, and that use case will keep growing [11,12]. However, treating API monetization as telecom's primary AI growth story confuses a useful feature with a business model.
For firms deciding where to place their next AI infrastructure bet, the diligence question is simple: are you buying access to a platform, or are you buying scarce physical capacity that no platform can substitute for? The operators and vendors who matter most over the next five years are the ones quietly turning unused power and fiber into AI infrastructure revenue, while the rest of the industry keeps debating which API aggregator will win. While it has not yet reached the headlines, this is a structural shift in the industry which is already directly impacting companies’ investment decisions and bottom lines.2
Sources
Figures cited are drawn from public company disclosures and third-party analyst estimates current as of mid-2026. Estimates vary across research providers due to differences in scope and methodology and are presented here for illustrative purposes. This article is for informational purposes only, and it does not constitute investment, financial, or legal advice.
BCE Inc., Q1 2026 Earnings Call, May 2026
TELUS, "TELUS Opens Canada's First Fully Sovereign AI Factory," media release, September 2025
SK Telecom, Q1 2026 Interim Report, May 2026
Ericsson, Q2 2025 Earnings Call, July 2025
Nokia, "Nokia Expands Network as Code Ecosystem," press release, March 2026
Light Reading, "Ericsson still in desperate hunt for network API profits," citing Omdia network API forecast data (also consistent with Juniper Research, IDC, and Analysys Mason forecasts)
McKinsey & Company, "What It Will Take for Telcos to Unlock Value From Network APIs," 2026
McKinsey & Company, "AI Infrastructure: A New Growth Avenue for Telco Operators," 2026
Juniper Research, Network API Revenue forecast, 2025 (low-end estimate, extrapolated from published 2025–2030 growth trajectory)
IDC, Worldwide Network API Forecast, 2025–2029 (IDC #US53143225), 2025 (high-end estimate, extrapolated from published 2025–2029 growth trajectory)
Omdia, "Telco B2B AI Monetization Index," July 2025
GSMA, "From Ambition to Execution: How Open Gateway Is Scaling the Global API Economy," 2026
Ericsson, Q1 2025 Earnings Call, April 2025
Verizon Communications Inc., Q2 2025 Earnings Call, July 2025