Industry Outlook: Telecoms & Connectivity — Week of July 20, 2026

Market Outlook

• AI‑native RAN moves from hype to scrutiny. Nokia is promoting what it calls the first AI‑native RAN platform, while analysts question how differentiated it really is and how fast operators can adopt it. In parallel, Omdia warns that regulators are still at early stages of concrete AI regulation, so operators face a mix of commercial pressure to add AI and policy uncertainty about how far automation can go in the network.
• US mobile speeds highlight 5G maturity gap. RootMetrics reports combined US nationwide mobile download speeds at 334 Mbit/s, with an increasing share of tests on 5G. The results show that dense, mid‑band heavy markets are now delivering on 5G marketing claims, which raises expectations for lagging markets and puts pressure on operators that have under‑invested in spectrum or densification.
• Regional operators refocus and consolidate bets. Telenor is exiting Asia to double down on the Nordics but still faces margin pressure at home, a pattern mirrored in other incumbents that are pruning portfolios to shore up balance sheets. Verizon is selling 274 retail stores and cutting another 500 corporate jobs, signaling a continued shift of investment away from legacy distribution and toward network and digital channels.

Discussion: CTOs should assume that board expectations on AI in the network will rise faster than regulators can provide clarity, and that 5G user experience benchmarks will harden into de facto competitive thresholds. Use this period to reset multi‑year capex roadmaps and be explicit about which assets are core and which are candidates for divestment or partnerships.

Headwinds

• Satellite D2D timelines slip and LEO awards shrink. AST SpaceMobile has pushed its direct‑to‑device satellite launch target to early 2027 and is seeking about 1 billion dollars to secure launch capacity and potential acquisitions. At the same time, US states are reportedly stripping up to 35 percent of locations originally awarded to LEO operators from subsidy programs based on updated mapping, which weakens the near‑term business case for some satellite broadband plays.
• Infrastructure friction threatens BEAD and rural builds. Electric co‑ops warn that FCC moves on pole attachment rules could slow BEAD‑funded builds, while utilities are already resisting new FCC rules meant to speed up pole access. Legal and procedural fights over poles risk delaying last‑mile fiber and fixed wireless projects just as political pressure for visible progress on rural broadband is intensifying.
• AI capex strain hits major vendors. Ericsson and IBM both reported Q2 pressure tied to AI infrastructure dynamics, where component costs and capex volatility are hitting margins. Vendor balance sheet stress can feed through into pricing, support quality, and R&D pacing for telco‑cloud, Open RAN, and 6G programs, especially for operators heavily tied to one or two strategic suppliers.

Discussion: CTOs should build more conservative timelines and dependency maps around satellite D2D and subsidized rural projects, and stress‑test vendor concentration risk under tighter balance sheet scenarios. Legal and regulatory friction around poles and spectrum should be treated as core program risks, not background noise.

Tailwinds

• US opens funding for AI‑enabled Open RAN. The NTIA is making up to 53 million dollars available to accelerate development, testing, and commercialization of secure, US‑led AI‑enabled RAN architectures. The program is explicitly aimed at Open RAN and AI in the radio domain, which gives operators and vendors a way to de‑risk early deployments and build domestic ecosystems around intelligent RAN functions.
• Edge and sovereign AI gain operator traction. Orange has expanded its own sovereign data center footprint to host a portfolio of AI models, anchoring telco‑controlled infrastructure as a home for sensitive data and low‑latency inference. Similar moves from other incumbents signal that telco‑cloud and edge sites are increasingly being positioned as AI infrastructure, not only as NFV platforms.
• 5G‑Advanced deployments showcase performance upside. China Unicom Beijing and Huawei have launched what they describe as the world's largest 5G‑A GigaUplink commercial network, highlighting uplink‑heavy use cases like live video and industrial telemetry. Such deployments set customer expectations for 5G‑Advanced features and will shape vendor roadmaps and spectrum strategies in other markets heading into the late 2020s.

Discussion: CTOs should treat new funding and early 5G‑Advanced networks as a signal to accelerate AI‑in‑RAN trials and to formalize an AI infrastructure role for telco edge. Early operator references will shape standards, ecosystems, and customer expectations, so decide where you want to lead versus follow.

Tech Implications

• AI‑native architectures reshape RAN and operations. Omdia highlights a shift toward autonomous capabilities in 6G‑era architectures, with operators layering AI into planning, optimization, and fault management rather than attempting blanket automation. Nokia's AI‑native RAN claims, plus NTIA's AI‑RAN funding, point toward a near future where policy‑driven, closed‑loop control, data pipelines, and model lifecycle management are first‑class design elements in RAN architectures.
• Telco‑cloud and sovereign AI converge at the edge. Orange's decision to host a range of AI models in its own data center shows how telco‑cloud stacks are being extended to support mixed NFV, containerized network functions, and AI workloads on shared infrastructure. That convergence raises questions around GPU scheduling, multi‑tenant isolation between network and AI workloads, and how to expose network APIs as native primitives to in‑house and third‑party models.
• Spectrum and satellite tensions shape 5G and 6G. SpaceX is lobbying the FCC for more flexible spectrum rules ahead of the next major auction, while a Zixi and Comcast Technology Solutions partnership offers an IP‑based alternative to upper C‑band video distribution as that band is repurposed. Combined with AST SpaceMobile's delay, the trend points to more contested mid‑band and S‑band allocations and a need to design RAN and core stacks for more dynamic spectrum sharing and interference scenarios involving LEO constellations.

Discussion: CTOs should push architecture teams to treat AI components, data pipelines, and spectrum agility as foundational, not bolt‑ons. That means concrete blueprints for AI‑enabled RAN, GPU‑aware telco‑cloud, and spectrum management that can cope with satellite coexistence and future 6G requirements.

CTO Action Items

Revisit your 3 to 5 year RAN and telco‑cloud roadmap with AI as a first‑order requirement, not a feature add‑on. Shortlist one or two AI‑in‑RAN or AI‑Ops pilots that can realistically tap NTIA or similar funding, and insist on clear data, model, and policy architectures. Ask your strategy and regulatory teams for a combined view of spectrum and satellite developments at the FCC and in your key markets, then bake more flexibility into band plans and radio procurement. Finally, pressure vendors on financial transparency and product roadmaps, and build at least one credible alternative path for any function that sits on a single strategic supplier today.