Operators Lean In On AI Grid Location

Earlier this week I argued that the AI Grid debate needs to move on from where you place a GPU to whether geographically dispersed compute can behave as a single fabric. I stand by that. But a story that has been building across the press this week is a useful reminder that the location question, the one I have been answering the same way for two years, is now being settled in public by the people who actually own the radio networks. And they are settling it against the tower.

The reporting is consistent. Light Reading describes Nokia and Nvidia's AI-RAN proposition running into telco resistance. Verizon, Vodafone, Orange and, notably for me, Telus have all raised doubts about putting graphics processing units into the radio access network. AT&T's chief technology officer has cast public doubt on the case for AI compute at the far edge. The enthusiasm for GPU-in-the-RAN comes from two operators, T-Mobile US and SoftBank, and almost no one else. Much of the rest of the industry is looking at Intel's newer CPUs for its open RAN rollouts rather than filling cell sites with accelerators.

I want to be precise about what this does and does not prove. It does not prove that AI in the RAN is a bad idea. Applying machine learning to scheduling, link adaptation and energy management inside the baseband is real, it is shipping, and Ericsson's AI-in-RAN software subscription is a reasonable way to bring it into existing hardware. What the operators are rejecting is narrower and more specific. They are rejecting the proposition that the cell site should become a general-purpose AI inference venue, stuffed with GPUs, monetised by hosting third-party workloads at the edge of the network. That is the proposition I have said for two years does not survive contact with power, cooling, space, security and, above all, the absence of a monetisation model.

My position has been that AI Grid deployment begins at the central office and the mobile switching office, not the cell site, because every physical and commercial constraint favours the aggregation point over the tower. The reasoning was never controversial to anyone who has stood in both kinds of building. A central office has power feeds, environmental control, physical security and fibre already in place. A cell site has a cabinet, a limited power budget and a landlord. When Verizon, Vodafone, Orange and Telus decline to put GPUs at the far edge, they are not making a new argument. They are confirming an old one, and they are confirming it with capital allocation decisions rather than conference slides, which is the only confirmation that counts.

There is a workstream reason this caught my eye. Telus appearing on the skeptics' list is consistent with what I see in the market: operators that are serious about autonomous operations are also the ones being disciplined about where AI compute physically lands. Those two forms of discipline are related. An operator that thinks clearly about the economics of edge inference tends to think clearly about the economics of everything else in the network.

The AI-RAN enthusiasm gap also matters for how we read vendor claims. When a technology has two vocal operator champions and a longer list of vocal operator skeptics, that is the signature of a capability that has been field-validated in specific conditions but not commercially validated across the market. I have made this distinction before and it applies cleanly here. SoftBank's agentic AI-RAN demonstrations and T-Mobile's Nvidia-backed edge trials are real engineering. They are not yet evidence that the model generalises to operators with different cost structures, different energy prices and different enterprise demand. Treat a two-operator enthusiasm as a pilot signal, not a market verdict.

So where does this leave the fabric argument I made last week? Exactly where I left it, and stronger. The operators are removing the least defensible node from the AI Grid, the cell site as inference host, which clears the ground for the argument that actually matters. Once you accept that heavy inference will not live at the tower, the interesting question becomes how you knit central offices, regional data centres and a small number of genuinely latency-bound edge sites into one addressable pool. The industry spent this week deciding where the compute will not go. That is progress. The harder decision, who owns the fabric that arbitrates across the places it will go, is still open, though.

AI Grid: Fabric vs Location Considerations

The debate about where AI compute belongs in a telecom network has been framed as a location question from the start. Do you put the GPUs at the cell site, the central office, the regional data centre, or the hyperscale campus? I have argued consistently that the honest answer begins at the central office and the mobile switching office, because power, cooling, fibre, physical security and latency sufficiency all favour those sites over the tower. That position has not changed. But two announcements from Asia this week suggest the more consequential question is no longer where the compute sits. It is whether the compute behaves as one pool regardless of where it sits.

NTT Docomo disclosed a nationwide testbed it calls GPU over APN. It pools graphics processing units spread across eight locations in five Japanese cities and presents them to a workload as a single platform, connected over the all-photonics network that NTT Group has been building under its IOWN programme. Docomo describes it as the realisation of its AI-Centric ICT Platform concept, part of what the group now labels AIOWN, its AI-native infrastructure. Strip away the acronyms and the claim is precise and significant: distributed GPUs, addressed as if co-located, over deterministic optical transport.

KT made the point from the other direction. Its new chief executive committed 18 trillion won, roughly 11.7 billion dollars, over three years, including 3.26 billion for one gigawatt of AI data centre capacity and a plan to connect that centralised infrastructure with edge sites serving low-latency workloads such as autonomous vehicles and industrial robotics. One operator is making dispersed compute act centralised. The other is extending centralised compute out to the edge. Both are describing the same thing from opposite ends, which is a compute fabric rather than a compute site.

This matters because it decouples two decisions the industry keeps conflating. Where you place a GPU is a question about power, land and cost. Where you run a workload is a question about latency, data gravity and sovereignty. As long as placement and execution are the same decision, every AI deployment becomes a real estate argument. Once a photonic fabric can make placement invisible to the workload, the two decisions separate. Training and heavy batch inference go where power and space are cheap. Latency-bound inference lands close to the user. The fabric arbitrates between them. This does not contradict the case for the central office, it absorbs it: the central office still wins for latency-bound edge inference, but that win is now a node in a graph rather than an isolated site.

I have some history with this problem. In 2018 I wrote about building at Telefonica what was probably the industry's first fully programmable multi-access edge computing platform, and the hardest part was never the compute. It was making distributed compute addressable, governable and billable as a coherent resource rather than a scatter of isolated sites. The technology around it has moved on considerably, but the unsolved problem is the same one Docomo is now attacking with photonics.

A practitioner's caution is in order. A fabric that makes national-scale GPUs behave as one pool is a testbed today, not a product. Docomo demonstrated it in a lab-grade programme. KT's edge connection is a plan, not a live deployment. Deterministic optical transport carrying commercial service level agreements under contended traffic is a materially harder thing than a controlled demonstration, and I would treat "as if co-located" the way I treat vendor energy savings figures: directionally real, quantitatively unproven at scale. The distance between a testbed that works and a fabric that carries production workloads is precisely the distance Open RAN spent five years crossing.

Still, the framing is the takeaway. The AI Grid conversation needs to move from siting to fabric. The operators that win will be the ones who can treat geographically dispersed compute as a single addressable resource, orchestrate workloads across it against real constraints, and price and settle access to it. That is a transport, orchestration and settlement problem before it is a property problem. The question is no longer which building holds the GPUs. It is who owns the fabric that makes the buildings irrelevant.

The Agent Runtime Is Not the Agent Model

DTW Ignite in Copenhagen made one thing clear: the vendor community has decided that the path to autonomous networks runs through agent runtimes. NVIDIA introduced NemoClaw blueprints and the OpenShell secure runtime to give long-running agents policy guardrails and sandboxed access to telecom systems. AdaptKey is piloting security-hardened agents for self-healing 5G operations. ServiceNow is bringing Project Arc to the NOC, orchestrating incident response from alert to work order. NTT DATA is building anomaly agents that escalate to research agents for telemetry analysis. Synthetic data rounds out the stack, a pragmatic answer to the fact that more than half of operators say their most valuable network data is too sensitive to use.

This is genuine progress and I do not want to minimize it. Containment, auditability and policy enforcement are necessary conditions for letting agents touch production networks. An agent that cannot be sandboxed cannot be trusted, and an agent whose actions cannot be audited cannot be certified. The runtime layer has to be built.

Containment is not coordination

But look carefully at what these announcements govern: individual agents, operating within a single operator's domain, executing workflows that a human has scoped in advance. This is vertical governance. It answers the question of whether an agent is allowed to perform an action. It does not answer the question that autonomous networks will actually pose at scale: when two agents are each permitted to act, and their permitted actions conflict, who decides?

Consider a scenario that is closer than most operators think. An enterprise logistics agent requests guaranteed throughput for a fleet of delivery robots. Simultaneously, a network energy agent, operating under its own perfectly valid mandate, is shutting down capacity in the same cluster to meet a sustainability target. Both agents are sandboxed. Both are auditable. Both are compliant with their policies. The runtime layer sees two well-behaved agents. The network sees a contradiction.

This is the problem I described in my previous post on network APIs. APIs were designed for developer access, not for agent-to-agent negotiation. Runtimes inherit the same blind spot. They secure the execution of each agent without providing any shared representation of the agentic plane itself.

What the meta-model requires

For agents to negotiate rather than collide, the industry needs a meta-model of the agentic plane: a topology of which agents exist and where they sit, an ontology so that an enterprise agent and a network agent mean the same thing by capacity, latency or priority, explicit authority boundaries defining what each agent may commit on behalf of its principal, shared state models so that negotiations reference the same view of the network, and audit trails that span negotiations rather than individual actions. None of the DTW announcements address this layer. They cannot, because it is not a product any single vendor can ship. It is a model the industry must agree on, the way it once agreed on network information models for OSS.

There is a familiar pattern here. The industry built firewalls before it built routing protocols for the internet's trust boundaries, and it spent two decades paying for the sequencing. We are building the firewalls of the agentic era first. The operators and standards bodies that formalize the agentic plane meta-model will define how enterprise AI and network AI transact for the next decade. The ones that stop at the runtime will discover that a network full of safely contained agents is not an autonomous network.

DTW Ignite 2026: The API Is Not Enough

I returned from DTW Ignite in Copenhagen with one conviction: the interface between enterprise applications and network infrastructure is about to change in a way the industry has not yet designed for.

Network APIs were never really about autonomous networks. That framing conflates two separate problems. APIs — CAMARA, GSMA Open Gateway, the decades of network exposure work that preceded them — were designed to let developers discover and consume network resources from outside the operator domain. Quality on Demand, location services, device status, number verification: clean REST interfaces exposed through a developer portal so that a programmer writing a B2B application could request a network capability and pay for it. Real progress on a real problem. But the problem was developer access, not network autonomy.

What is coming next is different in kind, not degree.

Enterprise AI agents are beginning to consume network infrastructure directly — not through a developer writing an integration, but autonomously, in real time, as part of executing a business objective. An industrial automation agent that needs guaranteed low-latency connectivity for a robotics fleet. A financial services agent that needs to provision a secure, isolated network path for a time-sensitive transaction. A logistics agent that needs to dynamically reserve bandwidth across multiple carrier domains as a shipment moves between jurisdictions. In none of these cases is there a developer in the loop. The agent has an intent, it needs network resources to fulfil it, and it needs to negotiate those resources with the network — now, at machine speed, without human mediation.

That negotiation cannot happen through a developer portal. It cannot happen through a static API catalogue with a PDF explaining what each endpoint does. The enterprise agent and the network need to speak to each other, and neither CAMARA nor MCP — whatever their respective merits — were designed for that conversation.

The network side of this exchange needs to be represented by network AI agents of its own: agents that can expose available capacity in real time, understand the constraints and commitments already in place, reason over competing demands, and negotiate resource allocation in a way that respects the network's operating boundaries. That is not a developer API. That is an autonomous counterparty.

And for those network agents to function — to negotiate reliably, to be governed, to be audited, to avoid conflicting with each other across RAN, transport, core, and the operational layers of OSS and BSS — they need something the industry is not yet building: a meta-model of the agentic plane itself.

Operators building autonomous networks are doing the right foundational work. Network topology models. Data ontologies. Decision layers. Closed-loop control architectures. These give the automation layer a complete and current picture of the environment it is operating in. But agents operating on that network need an equivalent model of themselves. Every agent with an identity, a capability scope, an authority boundary, a state, a dependency graph, and an audit trail. An abstract topology and ontology of agents, sitting alongside the topology and ontology of the network.

Without that model, what looks like autonomous negotiation between enterprise AI and network AI is actually uncontrolled interaction between systems that cannot see each other. An enterprise agent requesting bandwidth does not know what the network agent is authorised to commit. The network agent does not know what other network agents have already promised. No shared representation, no conflict detection, no governance.

The developer exposure problem is largely solved, or at least well understood. The agent-to-agent negotiation problem has barely been framed. That is the conversation the industry needs to have, and Copenhagen convinced me we are not having it yet.

O-RAN PlugFest Spring 2026: Smaller, Sharper, More Ambitious

The O-RAN ALLIANCE published results from its Global PlugFest Spring 2026 on June 4. Conducted from February to May 2026, it brought together 31 companies and institutions across 9 labs worldwide, co-hosted by 13 operators, OTICs and independent institutions. Those numbers will look like a step backward to anyone tracking PlugFest participation over time — and they are, deliberately.

For comparison: the Fall 2024 PlugFest ran across 28 labs with 115 participants. Spring 2025 had 19 labs and 69 participants. Spring 2026 cut the footprint by more than half again. This is not a loss of momentum. It is a deliberate shift from ecosystem-building to integration hardening. Getting everyone in the tent was the objective of the first five years. Making what exists actually work together is the objective now.

Massive MIMO — finally the main act

The centrepiece of Spring 2026 was multi-vendor end-to-end integration of massive MIMO O-RAN components, including mMIMO beamforming. This matters more than it might appear. Massive MIMO has always been the credibility gap for Open RAN. Traditional vendors have proprietary beamforming pipelines refined over a decade of commercial deployment. High-band, high-capacity urban coverage — stadiums, CBDs, transport hubs — has been essentially off-limits for disaggregated RAN. Demonstrating that multi-vendor mMIMO beamforming can be integrated in a structured, neutral lab environment is the first necessary step toward closing that gap.

AmpliTech's O-RAN CAT-B 64T64R Massive MIMO radio was the only radio of that configuration at the event — the only American-designed and commercialized O-RU at that specification level — and it demonstrated multi-vendor interoperability alongside operators including AT&T, Deutsche Telekom, Korea Telecom, LG Uplus, Orange and Rakuten Mobile. That combination — 64 transmit, 64 receive antennas, multi-vendor integration, neutral lab, named Tier-1 operator validation — is the kind of result that shifts a proof of concept into a procurement conversation.

Commercial-grade performance parity with incumbent vendors is a separate question, and a longer road. But you cannot get there without first doing what was done here.

AI-RAN — directionally consistent, details deferred

Several labs focused on AI-RAN solutions targeting network performance improvements and energy savings. The press release did not publish quantified results — a contrast with Spring 2025, which reported 25–30% intelligent RAN energy savings delivered by rApps on Non-Real-Time RICs. Detailed results will follow in technical session readouts. The direction is not in question. AI-driven energy efficiency via the RIC architecture has been the dominant AI-RAN use case in structured testing for two years running, and the logic is clear: energy is the largest controllable operating cost in RAN, the levers are cell on/off switching and power scaling, and the RIC is the right control point. What the Alliance needs to demonstrate next is that these savings hold at scale, under real traffic conditions, not just in lab scenarios with predictable load profiles.

The open-source stack is growing up

Four open-source frameworks were deployed at the PlugFest: Open Air Interface, OCUDU, O-RAN SC, and Sylva. Each represents a different layer of the stack, and their simultaneous presence in integration testing is worth noting.

OAI has become the de facto open-source O-CU/O-DU choice in lab settings — widely trusted, well-documented, increasingly operator-deployed in production pilots. O-RAN SC provides the RIC and SMO components from the Alliance's own software community. OCUDU is a combined CU/DU implementation targeting simplified deployment at the edge. And Sylva — the Linux Foundation's cloud-native telco infrastructure framework — is where the container orchestration and lifecycle management work happens. Carrier-grade Kubernetes for RAN workloads, in plain language.

The intersection of Sylva and O-RAN SC is the one to watch. Cloud infrastructure meeting RAN software — not as a research experiment, but as a jointly validated integration — is where operator confidence in virtualized open RAN will be won or lost operationally. It is still early. But it is no longer theoretical.

ISAC — a 6G signal in a 5G event

Initial tests were performed on Integrated Sensing and Communication, or ISAC — using the RAN simultaneously for wireless communications and environmental sensing. Object detection, positioning, radar-like environmental awareness, all from the same radio infrastructure. This is a 6G-era capability appearing in 5G Advanced specifications, and its inclusion in a 2026 PlugFest is a deliberate statement by the Alliance. They are not waiting for a new standards cycle to begin building the interoperability baseline. Given how long it took Open RAN to move from specification to commercial deployment, starting ISAC integration work in 2026 is not premature. It is probably necessary.

Who was there, and who wasn't

The full participant list is a useful read. Operators present included Korea Telecom, LG Uplus, Orange and Rakuten Mobile, with AT&T represented at the governance level through Brian Daly's TSC co-chairmanship. Academic institutions — Virginia Tech's Commonwealth Cyber Initiative, Iowa State University, North Carolina State University, ETRI and Fraunhofer HHI via the i14y Lab — signal that the research-to-commercialisation pipeline is being built at the institutional level, not just the vendor level. Software Radio Systems, the team behind srsRAN, continues to appear — their open-source 5G stack is quietly becoming a reference DU/CU implementation in operator testbeds.

What is absent is equally notable. No traditional vendors. The large incumbents engage with the Alliance at the specification level; they do not typically show up at PlugFest lab level. This is not surprising, but it is a structural tension. The integration maturity of disaggregated RAN will ultimately be measured in Tier-1 operator deployments, and those operators currently run incumbent infrastructure. The path from PlugFest to procurement runs through a comparison the incumbents are not participating in on neutral terms.

What this PlugFest actually means

The O-RAN Alliance has always faced a version of the same challenge: translating credible lab results into commercial deployments at scale. PlugFests prove interoperability in controlled conditions. Operators need it in live networks, under real traffic, with real SLAs and real interference environments. That gap has not closed. But the Spring 2026 agenda — massive MIMO beamforming, AI-RAN energy efficiency, integrated open-source stacks, ISAC first tests — reflects an ecosystem that knows precisely where it needs to focus to close it. That is not a small thing. Three years ago, the conversation was still largely about whether disaggregated RAN could work at all. That question has been answered. The question now is how fast it can reach performance and cost parity with incumbent solutions in high-capacity, high-stakes deployments. Spring 2026 moved that needle. Not dramatically. But in the right direction, on the right problems.

Innovating and Monetizing in Telecom: The Lean Telco

 

As alluded to in my previous posts, I have integrated the Lean Startup methodology, the design thinking framework and the Wardley Map model to create value in a telco environment.

Value is a subjective topic but in a Telco context, my efforts have been aimed at creating sustainable growth strategies. Very simply, sustainable growth comes from sustainable differentiation, which stems from the creation and evolution of technological, commercial and operational characteristics that become difficult, expensive and time consuming to emulate from your competition.

Sustainable growth comes from sustainable cost reduction and revenue growth (Duh!).

Sustainable cost reduction can be achieved through drastic cost structure changes. In 2026 Telco, it can be attained through the implementation of a cloud native architecture and principles, underpinned by strategies of network disaggregation, extensive use of open APIs and open network topologies; control / user plane separation and systematic automation. While these goals are challenging by themselves, particularly in a brownfield legacy telco environment, they are the bare necessary changes for survival. The challenges associated with the organization, skill sets and methodologies to evaluate, test, deploy, purchase and maintain these technologies are even larger.

Every telco is extremely skilled at managing technological and operational risk, through iterative, waterfall evaluation and tests, resulting in deployment of high availability and capacity networks. This methodology has also led to lengthy evaluation periods and deployments. Most vendor will recognize that the sales cycles in telco are over 2 years long and that making any change in a commercial network takes several million of dollars or euros. This has led to an oligopoly where only a handful of specialized vendors are able to sustain economically these drastic processes.

Lately, telcos have been trying to diversify the pool of vendors to increase competition and innovation by promoting open source and open API projects such as open RAN.

While these projects have shown interesting progress, the real cost reduction comes from the change in methodology and processes to take advantage of these more nimble vendors offering.

What I am proposing with Lean Telco is a methodological framework for identifying, evaluating, testing, sourcing, deploying telco products and services that will provide sustainable differentiation with drastically different cost structure than the incumbent versions.

Once you have successfully changed the cost structure of evaluating, buying, deploying and managing telco infrastructure and capacity, you can survive as a high capacity, low overhead provider of connectivity. But if you want to strive and grow, you need to attack the revenue part of the equation. Actually, one would argue should start with growth objectives, and look at cost structure as an optimization challenge.

Growing revenue sustainably, in a telco environment comes from either having more people using your existing services, or using more of them, connect new people or create new services. I have prototyped, tested and launched projects in each of these categories in my role at Telefonica.

1. Having more people use your existing products is difficult for telcos, because those products (residential and enterprise mobility, internet, telephony, TV...) are poorly differentiated, since they rely on the same technology from the same vendors. As a result most telcos end up trying to deploy first (5G, SDWAN...) or to claim a performance advantage, usually derived from a superior spectrum or infrastructure investment. The only real differentiation ends up being pricing. This is very expensive and not sustainable.
2. Having your customers using more of your service does not necessarily lead to more revenue, as bundles and unlimited plans are periodically rolled out to counter internet hyperscalers offering who rely on a different cost structure and revenue model. Again, since these services are mostly the same from one operator to another, differentiation comes from bundling and pricing. This is not sustainable.
3. Connecting new people / clients is a worthy endeavour, but the last unconnected live mostly in rural, low density areas and selling services to new corporate clients usually mean competing against public cloud offering that are more cost effective and flexible than what most telcos can offer. There are possibility of growth there, but it requires breaking out from the current telco technological framework and a willingness to assemble new value chains.
4. Creating new services is certainly where there is the most value, if we look at the growth of telephony over internet, video streaming services, social media and social messaging, SDWAN, cloud security, SASE, AIaaS... it is also the area with the most uncertainty and risk.

Telcos are not well equipped to manage the risk and uncertainty inherent in the discovery and creation of new services. The methodologies, organization and processes they use is to deliver with absolute certainty a product or utility with zero default to a mass market without variation. This model works well for mature, disciplined technology and vendors, not at all for exploration and innovation.

Too often, some Telcos build an extremely detailed plan, with contingencies. They budget it, staff it, resource it to execute it within a given timeframe, only to discover that the client didn't really want / need / value what was proposed (cf. push to talk, IMS/VoLTE, RCS, private networks...).

Just like in Lean Startup, the methodology I propose allows the progressive liberation of resource and funds as commercial uncertainty is shed by direct client interaction, testing and feedback. In a typical telco environment, the client interaction is at the very end of the process, here we are going to intersperse it throughout the development process to allow pivots, or early termination if the hypothesis are not met.

Trained, mentored and helped by many, I have adapted a few methodologies to enable Telcos to identify, validate, and deploy new services in an agile and cost effective fashion. I call it the Lean Telco Methodology.

How do I create a Lean Telco?

I use Wadley Maps for situational awareness and create a topographical representation of the current environment, which in my area of interest range from AI (sovereign, factories, grid, agentic, physical...), telco network cloud, orchestration, cloud native distribution and orchestration (K8, micro services) and hybrid cloud / edge computing (telco private stacks, AWS outpost, MS Azure, AI grid...). This is not a map until we apply the level of maturity (Genesis / handmade, Custom / solution, Product, Utility) to each of them, as well as their direction and barriers on the horizontal axis. On the vertical axis, instead of using Wardley's traditional visibility method, I use technology stacks such as access, transport, core network, OSS /BSS, orchestration... The purpose of the map is not to be precise or even right, it is to share and compare understanding of the environment, the players, their direction, velocity and the barriers. This visualization enables a level of shared understanding necessary to strategic discussions and gameplay around permutations and what-if? scenarios.

Once identified priorities and areas of risks / opportunities to investigate, I use the Lean UX framework and Lean Startup methodology to systematically identify potential current problems needing solving, unmet customer needs, unsatisfactory experiences and potential new products / services that customer wouldn't even know or have an opinion about. A series of workshop is usually best to crystalize the ideas. Once identified, they need to be refined into customer centric objectives. Design thinking frameworks help, but contrary to popular belief, customer centricity is not necessarily going to ask prospective customers about what they think. Most wouldn't have any idea about what to do with 6G, physical AI or a token exchange if you asked them. This is where lean UX and empathetic composite models are useful.

Each idea is reviewed by a jury and graded, the jury defines which ideas can make it to the next stage. The ideas are shaped and staffed as independent projects, with dedicated resource, budget and time box. Each project lead has the overall responsibility for moving the project to the next phase and to deliver the results of the current phase to justify additional resource and budget for the next one.

At a high level, the phases are:

• Ideation - ($5k-10k /1 - 3 months) -the idea is shaped into a project, with central opportunities, areas of innovation, right to play for the company, sustainable differentiating factor, commercial high level opportunity and cost / timing for the next phase.
• Prototyping - ($20k - 50k / 2 - 6 months) - In this phase a prototype is built, that might or might not incorporate any development or use of technology for the target invention. The idea is just to emulate the resolution of the problem and put it into customers hands as early as possible to identify whether the objectives, assumptions are framed properly and whether the client would value the resolution.
• Beta - ($300k - $600k / 3 - 6 months) -  once the central problems are identified, and we know the client values their resolution, it is time to create a MVP to prove that it is technically, commercially, organizationally possible to solve that problem and that the value created exceeds the costs.
• Product - ($1m - $3m / 3 - 6 months) - In this phase, once proven that the solution is possible, it is necessary to prove that the solution will scale and will be deployable with a mature operational and commercial model.
• Growth - (TBD) This is the phase where the project needs to be commercially and economically sustainable.

Each phase require client interactions, in the form of actual tests in conditions as close as possible to commercial network. Within each phase, we decompose the project into customer centric objectives. Each objective into hypothesis. Each hypothesis into series of experiments that will validate or invalidate the hypothesis. It helps to set clear expectations and success criteria for each of these.Wardley maps helps again, within each phase understanding what tasks, experiments are more suited for pioneers, settlers or town planners and indeed whether the project lead can adopt this mental posture in this phase or whether someone else needs to take the lead.

The result is a portfolio of new revenue making projects, that are systematically validated by customer feedback, capacity and propensity to pay; together with a robust operational and commercial model. Each project is periodically reviewed and graded, all projects must pass a gate review before the next phase and liberation of funds, which allow a nimble, measured, progressive investment plan, as risks and uncertainty decrease throughout the life of the project.

Non-RT RIC, AI-RAN, and the AI Grid: Three Different Bets on the Future of the RAN

I have been asked a few times lately what the difference is between the Non-Real Time RIC and AI-RAN. The question itself tells you something. Both sit under the broad "AI in the RAN" umbrella, marketed aggressively by the same vendors, debated in the same conference sessions. But they are fundamentally different in architecture, ambition, and business model. And neither is quite the same as what NVIDIA formally branded the AI Grid at GTC 2026 — which is where the most important and most misread opportunity actually sits.

The Non-RT RIC: the pragmatic bet

The Non-RT RIC is an O-RAN defined software layer that sits in the Service Management and Orchestration layer above the RAN, not inside it. Control loops over one second. rApps for energy saving, traffic steering, slice assurance, automated optimization. Think of it as the evolution of Self-Organizing Networks, re-platformed on open interfaces with a proper application model and a genuinely lower barrier to entry — cloud-native and OSS skills are sufficient. No RAN silicon expertise required.

This is precisely why the early commercial traction is here, not in AI-RAN. AT&T is deploying Ericsson's SMO and Non-RT RIC to replace two legacy C-SON systems. TELUS has launched an RIC platform alongside its Open RAN rollout. Swisscom is deploying one for multi-technology network management. These are not trials. These are production decisions.

AI-RAN: real performance gains, speculative revenue

AI-RAN embeds AI natively into the RAN stack itself .The AI-RAN Alliance — founded in February 2024, now at 109 member companies — defines it across three working groups: AI-for-RAN, AI-and-RAN, and AI-on-RAN.

AI-for-RAN is the most mature: using AI to optimize the RAN itself — the scheduler, link adaptation, beamforming, interference management. T-Mobile and Ericsson have been trialing an AI-driven scheduler and link adaptation engine on a live 5G Advanced network since Q2 2025, targeting commercial deployment in Q3 2026. Nokia and NVIDIA, backed by a $1 billion equity partnership, are testing GPU-accelerated AI-RAN with BT, Elisa, NTT DOCOMO, and Vodafone.

AI-and-RAN is where the narrative gets more ambitious — and more speculative. The idea is that RAN sites become shared compute infrastructure, running both network workloads and enterprise AI workloads on the same hardware. The tower becomes a distributed AI compute node. New revenue streams. Operators escape the utility trap.

AI-on-RAN is the monetization layer for the above. The commercial mechanisms are still being defined. That tells you where the maturity is.

The AI Grid: follow NVIDIA's sequencing, not its marketing

At GTC 2026, NVIDIA formally introduced the AI Grid as a reference design — geographically distributed AI infrastructure, using the telco footprint to run inference workloads closer to users. The numbers are interesting: early Comcast benchmarks showed inference cost reductions of up to 76% versus centralized deployments. HPE, SpectroCloud, and others have already announced implementations aligned to the reference architecture.

I have used this concept in my own work for years to describe the evolution from isolated MEC deployments into a coherent, programmable distributed inference fabric. Good to see NVIDIA put a formal architecture behind it. But the marketing obscures a critical sequencing question.

NVIDIA's own GTC announcements noted that many operators are starting by lighting up existing wired edge sites — central offices and mobile switching offices — as AI Grids they can monetize today. The cell site layer is a later phase. AT&T's CTO Igal Elbaz has been direct about questioning the value of pushing compute all the way to the far edge to save one or two milliseconds of latency. T-Mobile's SVP of network infrastructure defined her AI edge strategy as what is at a data center at a mobile switching office. Verizon's CTO has flagged the cost and complexity of far-edge GPU deployments.

These are the three largest US operators. They are not being conservative for the sake of it. The economics are straightforward: central offices and mobile switching offices already have power, cooling, connectivity, and physical security. They aggregate traffic from hundreds of cell sites. The sub-500ms latency threshold that NVIDIA's own reference design targets is achievable from a well-positioned CO. It does not require a GPU at the tower — not for the use cases that have a business case today.

I have seen this movie before with MEC. The industry led with its most ambitious architectural vision, ran the infrastructure investment ahead of the demand, and recovered slowly. The AI Grid does not have to repeat that pattern.

What to actually do

Start with the Non-RT RIC. The contracts are being signed, the ecosystem is opening, the business case is defensible.

On AI-RAN, wait for AI-for-RAN where your vendors have credible near-term roadmaps. Treat AI-and-RAN at the cell site as a long term speculative option — worth tracking, too early to fund at scale.

On the AI Grid, follow NVIDIA's own sequencing rather than the brochure. Central offices and mobile switching offices first. Build the orchestration and service layer from there outward. Expand to the far edge when the use cases and economics justify it — not because a GPU manufacturer's demand forecast requires it.

The cell site AI Grid is a compelling long-term vision. The central office AI Grid is deployable today. In this industry, deployable usually wins.

From AI-Native to Agentic-Native Networks

Recent announcements at NVIDIA's GTC 2026—including major pushes into agentic AI frameworks like OpenClaw, NemoClaw, and agentic systems for reasoning, planning, and autonomous action—have reinforced several convictions I've held about the trajectory of AI-native infrastructure, especially in telecom and networked industries. We're seeing the emergence of two distinct paradigms:

• AI-Native networks that observe, detect, optimize, and predict in real time. These systems augment human decision-making, providing powerful assistance in planning, deploying, and managing both physical and virtual infrastructure.
• Agentic-Native networks, by contrast, eliminate the human-in-the-loop entirely. When equipped with real-time data access, transactional capabilities, and fulfillment capacity, they execute at the speed permitted only by the slowest link in the supply chain.

This second model doesn't just accelerate execution—it fundamentally reprices time itself as a competitive asset.

As Jordi Visser articulates in his insightful piece "The Repricing of Time: Equity in the Age of Agents", agentic AI compresses competitive cycles dramatically. Velocity of execution no longer merely helps fulfill a plan faster; it redefines the playing field. When capabilities can be reconfigured almost overnight through model iterations or agent orchestration, durable moats erode. What once took decades to build—layered expertise, entrenched positions, regulatory barriers—can now be challenged or leapfrogged in months.

In this environment, equity behaves more like a call option on execution speed than on long-duration stability. "Execution speed replaces installed base. Iteration cadence replaces headcount." The advantage shifts decisively toward those who can pivot, adapt, innovate, and execute rapidly.

This dynamic hits telecom particularly hard.

Most operators are desperate to escape the "utility trench"—the low-margin, commodity perception that has trapped connectivity providers for years. They aspire to new revenue streams beyond pipes and bandwidth.

From my own experience modeling, teaching, and advising organizations on this challenge (see earlier pieces on innovation micro-strategies, telco relevance and growth, and the lean telco), there is no single silver bullet. No grand transformation program that magically reinvents the business.

Instead, the path forward involves thousands of micro-services and experiments: create, test, fail fast, pivot, scale the winners, and launch repeatedly. The era of one-size-fits-all offerings is over.

Agentic-native networks offer exactly the infrastructure to make this high-velocity approach viable at scale. They enable rapid creation, iteration, value capture, and deployment—turning velocity, flawless execution, and clear strategic vision into the new currency that outcompetes inertia, legacy systems, and eroding differentiation.

For telecom leaders, the message from GTC 2026 is clear: agentic AI can free up resources and help accelerate innovation at scale. Those who embrace this shift—building or partnering for agentic capabilities—will be the ones that don't just survive the repricing of time, but help define the next era of networked value creation.