Cannes, Two Sessions, One Direction.
Data Cloud Global Congress 2026 brought the usual density of announcements, corridor conversations, and competing narratives about the future of AI infrastructure. I left Cannes more certain than ever that the sector is approaching a genuine inflection point, not just on power availability, but on how the industry chooses to think about the relationship between speed, resilience, and decarbonization.
I had the privilege of contributing to two sessions at this year's event, and while they addressed different aspects of the power challenge, both pointed to the same underlying conclusion: the energy transition for AI data centers is not a binary choice between "power now" and "power clean." Structured transition is possible. Renewable natural gas (RNG), or biomethane; cleaned biogas, sits at the heart of that argument.
The Structured Transition Model
This session, delivered on the Engage Stage on June 3, marked the public launch of A Structured Transition Model for AI Data Center Power, which is now hosted on a dedicated Rehlko microsite as the primary publication location.
Launching the Structured Transition Model with Francis Perrin, CMO of Rehlko and Miranda Gardiner, iMasons
The structured transition model is a response to a pattern I have observed repeatedly across hyperscaler, colocation, and industrial power conversations over the past several years. Operators face a trilemma: they need power fast, they need it to be reliable, and they need it to be on a credible decarbonization trajectory. No single technology currently satisfies all three simultaneously. But a structured, phased approach can.
The model sets out a framework in which distributed gas generation, potentially fueled by RNG or hydrogen, provides the immediate power foundation, grid integration is pursued in parallel as capacity becomes available, and progressive decarbonization options (hydrogen-ready engines, carbon capture, renewable fuel switching) are built in as optionality rather than retrofitted as obligation. The key principle is that decisions made under speed-to-power pressure do not have to foreclose future sustainability outcomes, if the architecture is designed with transition in mind from the outset.
This is not an abstract concern. In Europe in particular, the builders I spoke with are already thinking past the speed-to-power phase. They are asking what happens to these gas assets once grid capacity eventually arrives: whether the engines can be redeployed, switched to a different fuel, or repurposed for resilience and grid services rather than left stranded. That question is no longer a footnote in the planning conversation. It is shaping procurement decisions today, which is precisely the behavior the structured transition model is built to support.
Launching this model at Datacloud felt appropriate. The audience in Cannes, operators, developers, investors, and infrastructure providers, is precisely the group that will determine whether the AI infrastructure build-out of the next decade happens in a way that is consistent with long-term climate commitments, or whether the urgency of today's demand creates a lock-in that is difficult to unwind.
It is also worth noting that onsite power with gas engines can deliver carbon benefits even compared to France's local electricity grid, which has high nuclear penetration. The carbon case for distributed generation is more nuanced than the grid-versus-generation framing often suggests.
Part of that nuance is what you do with the heat. Combined cooling and power remains, to my mind, one of the most under-appreciated levers in the entire data center energy conversation. A gas engine rejects a large share of its input energy as heat, and in a conventional setup that heat is simply lost. Data centers, however, carry an enormous and continuous cooling demand, and that recovered heat can be put to work driving absorption cooling rather than wasted. Measure carbon not per unit of fuel burned but per useful unit of energy delivered, including the cooling the facility would otherwise have to generate separately, and the efficiency case shifts considerably. We spend a great deal of time debating fuels and not nearly enough debating how completely we use the energy once it is in the building.
Session Two: From Grid Delay to Power Today - Rethinking Green Energy
The second session, hosted by Future Biogas also on the Engage Stage, opened with a question the whole industry is grappling with: how do you access power in months when the grid is offering you years?
Future Biogas Session with Chris Huhne, Former Secretary of State for Energy and Climate Change, and Suneel Appan, Sector Lead - Data Centres and CCUS, Future Biogas
Grid connection delays across Europe and North America are not theoretical. They are active constraints on investment decisions, deployment timelines, and AI infrastructure scaling. Multi-year queues for grid capacity are forcing developers to make uncomfortable choices, and in many cases, those choices are defaulting to diesel and fossil gas at scale, with the sustainability implications simply deferred.
And this is not a problem confined to a handful of constrained markets. The grid is under strain in most parts of the world, and the incremental loads that data centers are now deploying are larger than what the grid can supply on the timescales these projects demand. The mismatch is not at the margins. A single large campus can request more power than a mid-sized town, and it wants that power in quarters rather than decades. The queue, in that sense, is a symptom. The underlying issue is that demand has outrun the rate at which transmission and generation can physically be added.
The session explored why that trade-off is a false one.
Biogas, and specifically RNG produced from anaerobic digestion of organic waste, represents a genuinely different category of solution. Let me be precise about why, because the distinction matters.
RNG is not a carbon credit. It is a fuel.
This is the first and most important point. When a data center procures RNG and uses it to power on-site generation, it is combusting a fuel whose carbon was recently captured from the atmosphere as part of biological cycles, not released from ancient geological reserves. Unlike purchasing Renewable Energy Certificates or carbon offsets, which are accounting mechanisms that match consumption with renewable generation somewhere in the system, RNG can be consumed directly and in real time. The emissions offset is not hypothetical or spatially distant. It is occurring at the point of generation. For operators under pressure to demonstrate authentic, verifiable emissions reductions rather than creative accounting, this distinction is significant.
The timing advantage of RNG is structural, not incidental.
Biogas-fueled distributed generation can be deployed in months using existing gas infrastructure, established engine technology, and proven project financing models. Energy-as-a-service structures remove the need for operators to own the generation asset, lowering capex exposure while maintaining resilience. This is not a niche or emerging configuration. It is a bankable, scalable model that can reach financial close on timelines that grid connections simply cannot match.
Coupled with CO2 capture, RNG generation becomes net carbon negative.
This is perhaps the most underappreciated aspect of the biogas opportunity for data centers. The combustion of RNG in a gas engine is already broadly carbon-neutral on a lifecycle basis. Add post-combustion carbon capture, capturing the CO2 from the exhaust stream, and the energy system transitions from neutral to negative. The data center becomes not just a clean consumer of power, but an active participant in carbon removal. For operators navigating Scope 1, 2, and 3 commitments, the long-term potential of this configuration is material.
The Demand Anchor Argument
One further point that generated real discussion in the biogas session deserves amplification here.
The World Biogas Association has set out a clear target: RNG has the potential to contribute an 11% reduction in global greenhouse gas emissions by 2030. That is not a marginal contribution. It is sector-transforming at scale. But realizing that potential requires production to scale significantly beyond current levels.
Data centers, as large and predictable energy consumers, have an unusual opportunity here. By committing to RNG offtake, even partially, even as a blended fuel strategy, they can act as demand anchors that make new biogas production projects financially viable. This is the same logic that has driven corporate PPA markets for wind and solar over the past decade: large buyers creating the demand certainty that unlocks project finance and stimulates supply. The difference with biogas is that the fuel is dispatchable, storable, and can be consumed when and where it is needed, properties that solar and wind cannot offer without significant storage infrastructure.
If even a fraction of the AI data center capacity being planned over the next five years committed to RNG procurement as part of a structured transition strategy, the signal to the biogas production sector would be transformative. Supply would follow demand. The 2030 targets become more achievable. And the data centers themselves would be positioned with a credible, demonstrable decarbonization story at a time when scrutiny of the sector's environmental footprint is intensifying.
What I Took Away from Cannes
The conversations in Cannes confirmed several things I had suspected but now feel with more conviction.
The industry is genuinely searching for solutions that work across the full trilemma, not just power fast, but power fast and credibly. The appetite for practical, bankable, near-term alternatives to diesel and fossil gas is real. And the biogas sector, long considered a specialist or agricultural story, is increasingly being taken seriously as a mainstream energy infrastructure solution for digital infrastructure.
One signal stood out in almost every conversation I had. Reputable companies that can credibly supply large blocks of power, north of 20MW, with delivery in 2027 and 2028, are in remarkable demand. The speed-to-power dilemma has become concrete and time-bound. It is no longer a question of whether the industry takes near-term generation seriously, but of who can actually deliver at scale and on a date that matters. Capacity with a credible delivery window is now one of the scarcest commodities in the sector.
A smaller observation, but one I keep returning to. Cannes this year drew large contingents from the UK, Ireland, the United States, and Israel. The domestic French presence felt noticeably thinner than I had expected, given the venue. I would not read too much into a single event, but it does say something about where the urgency around AI infrastructure is concentrated right now, and perhaps about which markets feel the speed-to-power pressure most acutely.
The structured transition model is not a perfect answer. No single framework can accommodate the full diversity of operator circumstances, grid environments, regulatory contexts, and decarbonization timelines. But it is a rigorous attempt to provide a decision-making architecture that holds all three imperatives in tension, and that is deployable today, not dependent on technologies that are still five years from commercialization.
If you are working through these questions in your own organization, whether as an operator, developer, investor, or policy actor, I would welcome the conversation.
The full model is available via Rehlko and the thinking behind it here. I will also be publishing further commentary on individual components of the framework over the coming months.
What was your key takeaway from Data Cloud Global Congress 2026?
What is the Structured Transition Model?
The Structured Transition Model is a framework designed to help data center operators navigate what is increasingly described as a three-part challenge. They need power immediately to support demand, they need it to be reliable, and they need to ensure it aligns with decarbonization commitments. At present, no single technology delivers all three.
Rather than forcing a binary choice, the model sets out a phased approach. It begins with distributed generation, often using gas engines that can be deployed quickly. At the same time, grid connections are pursued as capacity becomes available. Over time, the system is designed to evolve, incorporating cleaner fuels such as renewable natural gas or hydrogen, and potentially adding carbon capture. The central idea is that early decisions do not have to limit future sustainability outcomes if the system is designed with transition in mind from the outset. The thinking behind the STM can be found here and the background to the Structured Transition Model is explored in the book Five Nines and Fast Power.
Why is this model especially relevant now?
Its relevance comes directly from the reality of grid delays. In many parts of Europe and North America, developers are facing multi-year waits for grid connections. This is no longer a theoretical issue; it is actively shaping project timelines and investment decisions.
As a result, many operators are defaulting to diesel or conventional fossil gas simply to get projects moving, with sustainability considerations postponed. What the Structured Transition Model demonstrates is that this trade-off is not unavoidable. It is possible to secure power in the near term while still building toward a lower-carbon future, provided the system is designed properly from the beginning.
The grid delays you mention, how severe is the mismatch really?
It is more fundamental than a queue. The grid is under strain in most parts of the world, and the incremental loads that data centers are deploying are larger than what the system can supply on the timescales these projects need. A single large campus can request more power than a mid-sized town, and it wants that power in quarters rather than decades. So the multi-year connection queues are a symptom rather than the root problem. Demand has outrun the rate at which transmission and generation can physically be added. That is why near-term, deployable generation has moved from a contingency to a central part of the conversation.
If operators build gas generation now, are they not just creating stranded assets later?
This is exactly the question I heard most often in Europe, and it is the right one to ask. The concern is legitimate if the generation is treated as a one-off fix. It is far less of a concern if the asset is designed for a second life from the outset. Engines can be redeployed, switched to renewable natural gas or hydrogen as those fuels become available, or repurposed for resilience and grid services once grid capacity arrives. The point of designing for transition is that the decision you make under speed-to-power pressure does not lock you into a dead end. Builders are increasingly making procurement choices on that basis now, rather than deferring the question.
What role does renewable natural gas play in this approach?
Renewable natural gas, or biomethane, sits at the center of the model because it provides a bridge between immediate operational needs and long-term decarbonization goals. It behaves like conventional gas in terms of reliability and dispatchability, but its carbon characteristics are fundamentally different.
Because it is derived from organic waste, the carbon released during combustion is part of a relatively recent biological cycle rather than being drawn from fossil reserves. This allows operators to deploy familiar technologies while materially improving their emissions profile. Crucially, it offers a pathway to cleaner operation without requiring a complete redesign of existing infrastructure.
How is RNG different from carbon credits or renewable certificates?
The distinction is important. Renewable natural gas is a physical fuel that is consumed directly at the point of generation. The emissions profile is tied to what is actually happening on site, in real time.
By contrast, carbon credits and renewable energy certificates are accounting mechanisms. They represent emissions reductions or renewable generation that may occur elsewhere and at a different time. For organizations under increasing scrutiny to demonstrate genuine and verifiable emissions reductions, this difference carries significant weight. RNG provides a more direct and transparent link between energy use and environmental impact.
Can RNG-based systems be deployed quickly enough?
One of the most compelling aspects of RNG is the speed at which it can be implemented. Because it relies on established engine technologies and existing gas infrastructure, projects can typically be delivered in months rather than years. Financing structures are also well understood, including energy-as-a-service models that reduce upfront capital requirements.
This combination of technical maturity and financial viability makes RNG-based distributed generation a practical solution for bridging current power constraints without delaying development timelines.
What is the market actually short of right now?
Credible, near-term capacity. The clearest signal from almost every conversation I had in Cannes was demand for reputable companies that can supply large blocks of power, north of 20MW, with delivery in 2027 and 2028. The speed-to-power dilemma has become concrete and time-bound. It is no longer about whether the industry takes near-term generation seriously, but about who can actually deliver at scale on a date that matters. Capacity with a credible delivery window is now one of the scarcest commodities in the sector.
Can RNG go beyond carbon neutrality?
When combined with carbon capture, it can. On its own, RNG is generally considered carbon-neutral on a lifecycle basis. However, if the carbon dioxide produced during combustion is captured and stored, the system can effectively remove carbon from the atmosphere.
This creates the possibility for data centers to move beyond simply reducing their impact and toward playing an active role in carbon removal. Over time, this could become a significant consideration for operators managing increasingly complex emissions commitments across Scope 1, 2, and 3.
Where does combined cooling and power fit into this?
It is one of the most under-appreciated levers in the entire data center energy conversation. A gas engine rejects a large share of its input energy as heat, and in a conventional setup that heat is wasted. Data centers carry an enormous and continuous cooling demand, and that recovered heat can be used to drive absorption cooling instead. If you measure carbon per useful unit of energy delivered, including the cooling the facility would otherwise have to generate separately, rather than per unit of fuel burned, the efficiency case improves considerably. We spend a great deal of time debating fuels and not nearly enough on how completely we use the energy once it is in the building.
What is the "demand anchor" argument?
The demand anchor concept reflects the role that large, consistent energy users can play in scaling new energy industries. The World Biogas Association has identified significant global emissions reduction potential from biogas, but reaching that scale requires increased production capacity.
Data centers are uniquely positioned to support this growth. By committing to long-term RNG offtake, they can provide the demand certainty needed to unlock new projects and attract investment. This is similar to how corporate power purchase agreements accelerated the growth of wind and solar markets. The difference is that RNG offers dispatchability and storage, making it particularly well suited to the needs of digital infrastructure.
Why is RNG well suited to data centers specifically?
Data centers require constant, reliable power, and intermittent renewables alone cannot meet that requirement without substantial storage. RNG provides a complementary solution because it is dispatchable and controllable, meaning power can be generated exactly when it is needed.
It also integrates easily with existing infrastructure, reducing complexity and deployment risk. These characteristics make it highly aligned with the operational demands of AI-driven data centers, where uptime and resilience are critical.
What did the conversations in Cannes reveal?
There was a noticeable shift in tone compared to previous years. The urgency around power availability is now widely accepted, but so too is the need for credible sustainability strategies. Operators are actively looking for solutions that can deliver both.
At the same time, the perception of biogas is changing. What was once seen as a niche or agricultural solution is increasingly being recognized as a viable component of mainstream energy infrastructure, particularly for distributed generation.
Did the makeup of the event tell you anything?
A smaller observation, but one I keep returning to. Cannes this year drew large contingents from the UK, Ireland, the United States, and Israel, while the domestic French presence felt thinner than I had expected given the venue. I would not over-interpret a single event, but it does say something about where the urgency around AI infrastructure is concentrated at the moment, and perhaps about which markets feel the speed-to-power pressure most acutely.
Is the Structured Transition Model a universal solution?
It is not intended to be a one-size-fits-all answer. Different regions, regulatory environments, and operational priorities mean that no single approach will suit every case. What the model provides is a structured way of thinking about complex decisions, helping organizations balance competing demands while preserving flexibility for the future.
What should industry stakeholders do next?
The immediate priority is to evaluate how a structured transition approach could work within their own projects. This includes looking at how on-site generation can be integrated with future grid access, exploring fuel strategies that include RNG, and designing systems that can evolve over time.
The underlying message is simple: act with urgency, but design with foresight. Decisions made today will shape not just how quickly infrastructure is delivered, but how sustainable it ultimately becomes.
Where can I learn more?
The full Structured Transition Model is available via Rehlko, with further detailed commentary on individual components to follow. As the industry continues to evolve, ongoing dialogue between operators, developers, investors, and policymakers will be essential in shaping a balanced and effective path forward. The thinking behind the STM can be found here and the background to the Structured Transition Model is explored in the book Five Nines and Fast Power.