Mark A. Cleary

Technology Investor, AI Focused. $NVDA, $IREN, $CIFR

[Mark A. Cleary @MarkACleary2]

Amazing work. I agree with your thesis completely.

[franklee6924x @franklee6924T]

NVIDIA DSX System: IREN's Commercial Potential Under the DSX AI Factory Model (Part 3)

Before discussing this topic, I want to first analyze IREN's position within the DSX ecosystem.

According to NVIDIA's own description, under the DSX framework, it is collaborating with eight AI infrastructure companies. Besides IREN, the other seven are CoreWeave, Crusoe, Nscale, Lambda, Nebius, Firmus, and Yotta.

After reviewing the publicly available information regarding each partnership, the collaboration details can be summarized as follows:

CoreWeave — Uses NVIDIA DSX Air to build and test AI Factory digital twins in the cloud, allowing operational simulations before physical equipment is delivered and shortening validation cycles. It is also one of the cloud partners deploying DSX platform components, including DSX Sim, MaxLPS, and DSX OS.

Crusoe — Since March 2026, it has expanded its full-stack collaboration with NVIDIA, adopting the Vera Rubin DSX reference design and Omniverse DSX Blueprint to plan and operate gigawatt-scale AI factories. The partnership covers digital twins, AI-driven power and cooling optimization, and mechanical and electrical engineering design. Based on current information, Crusoe is the only company besides IREN that has a dedicated public announcement regarding gigawatt-scale physical engineering collaboration. However, NVIDIA has not designated Crusoe as a "flagship" partner.

Nscale — Not only is it one of the co-developers of the Vera Rubin DSX reference design itself, alongside engineering and simulation companies such as Cadence, Eaton, Jacobs, Schneider Electric, Siemens, and Vertiv, but it is also one of the eight cloud partners deploying DSX software platform components. Therefore, Nscale appears to have a certain level of involvement on the engineering design side.

Firmus, Lambda, Nebius, and Yotta Data Services — Public information on these four companies largely remains limited to a common press release description: as NVIDIA cloud partners, they deploy the core DSX platform components, including DSX Sim, DSX MaxLPS, and DSX OS, to reduce risk, improve GPU utilization, and accelerate infrastructure deployment. No dedicated announcements have been found indicating deeper physical engineering collaboration.

Among these seven companies, only Crusoe appears to have engineering-related cooperation. The rest are primarily involved in specific software-layer integrations, deployments, or operational applications.

By comparison, NVIDIA's collaboration with IREN demonstrates several unique characteristics in both depth and breadth.

First, logical inference suggests that IREN was likely involved during the initial planning and design stages.

The engineering component represents the overwhelming majority of the DSX system. A design framework of this scale must be built around a real, physically deployable site. Without a concrete location, it is impossible to create a truly executable project plan. Engineers must repeatedly validate boundary conditions such as power capacity, land availability, and cooling requirements against an actual site. This is basic common sense in complex systems engineering.

In November 2024, equipment procurement had already begun for IREN's 1.4GW Sweetwater campus in Texas.

A system design project as large as DSX spans multiple engineering disciplines and requires extensive feasibility verification against real-world infrastructure. Such a reference architecture would likely require at least one to one-and-a-half years of preparation.

By the time DSX was officially announced in March 2026, the overall framework was already largely complete, and the project had reached the stage where a flagship AI factory was ready for deployment.

Working backward from that date, the timeline aligns closely.

As the only known gigawatt-scale site in the United States over the past two years with a confirmed power energization schedule, Sweetwater has never publicly announced a specific monetization plan. To this day, no customer has been formally attached to the site. The only disclosed information is that it will serve as the flagship deployment site for the DSX AI Factory system.

The other company involved in DSX engineering collaboration, Crusoe, appears to have a significantly different relationship with NVIDIA.

While Crusoe also controls gigawatt-scale infrastructure, its sites already have clearly defined end users. The 1.2GW Abilene campus is leased to Oracle and OpenAI. Crusoe's business model with these customers is itself an innovative undertaking.

Because these facilities already have committed users and contractual obligations, it seems unlikely that they would have been available from the outset as dedicated engineering testbeds for NVIDIA's DSX development efforts.

As a result, NVIDIA's engineering collaboration with Crusoe is more likely focused on specific project phases or selected engineering disciplines.

Another important distinction is that Crusoe is a leading representative of the Behind-the-Meter (BTM) power generation model. Different power architectures create different engineering requirements. Therefore, the NVIDIA-Crusoe collaboration may be more focused on BTM-specific technologies.

BTM data center infrastructure remains a relatively new industry segment. Significant challenges still exist regarding reliability and regulatory compliance. Crusoe's history of project delays and cancellations reflects some of these realities.

Second, IREN's $3.4 billion services agreement with NVIDIA is unique.

Based on currently available information, no other DSX cloud partner has been disclosed as having a similar direct paid customer relationship with NVIDIA.

NVIDIA is effectively outsourcing part of its internal AI research workloads to an external operator. This arrangement appears to be unique within the DSX partner ecosystem.

This also indirectly demonstrates both the depth and duration of the relationship.

The deployment location for this agreement is IREN's Childress campus.

Childress combines high-capacity fiber connectivity, liquid cooling infrastructure, high-density rack deployments, and both air-cooled and liquid-cooled environments. Across virtually every infrastructure category, it represents a best-in-class configuration.

It is also arguably the most complete single-site environment for designing and validating the supporting components required by the DSX AI Factory architecture.

Again, this appears to be unique to IREN.

Putting all these factors together, Sweetwater's long-standing strategic ambiguity, combined with the recent announcement identifying it as the flagship DSX AI Factory site, suggests that cooperation between NVIDIA and IREN likely began at the earliest design stages.

NVIDIA's direct paid customer relationship with IREN is unique among all eight partners.

The flagship AI Factory under the DSX framework is being co-developed with IREN.

A flagship designation fundamentally implies uniqueness, exclusivity, and top priority.

Its deeper significance is that IREN gains earlier access to NVIDIA's core DSX architectural concepts and technical standards.

This has major implications for planning the future development of IREN's 5.8GW pipeline of secured power capacity.

It also reinforces an important lesson in this industry:

Being first is not what matters most.

Getting it right is.

Patience can create enormous value.

So what ultimately defines success for the DSX system?

In my view, it can be summarized in one sentence:

Lowest cost, highest output.

Jensen Huang has repeatedly expressed a similar objective:

"Maximize token performance per megawatt at the lowest token cost."

Anyone familiar with my long-term analysis of IREN will immediately recognize a striking alignment.

$IREN's operational philosophy has always been:

Lowest cost, highest output.

For NVIDIA, $IREN may represent the ideal partner for achieving this objective.

This helps explain why IREN alone has received flagship partner status.

From a broader commercial perspective, NVIDIA will certainly not limit itself to working exclusively with $IREN.

The DSX system will likely expand commercially across all seven remaining partners.

However, the amount of value ultimately created by each partner will depend on its own business model and execution capabilities.

NVIDIA undoubtedly hopes all partners succeed.

Yet there remains only one ultimate metric:

Lowest cost, highest output.

Whether it is called an AI Factory or a Token Factory is largely irrelevant.

If you achieve the lowest cost and highest output, you become the industry's dominant leader.

Naturally, NVIDIA's DSX resources will gravitate toward the highest-performing operators.

This is inevitable because it reinforces NVIDIA's leadership position across the broader AI ecosystem.

Once the lowest-cost, highest-output model is achieved, IREN's commercial opportunity could change dramatically.

Today, much of the market still evaluates AI infrastructure through an internet-era lens, assuming that software stacks are the primary value anchor of AI cloud businesses.

In reality, software moats are gradually weakening.

The future benchmark for AI cloud infrastructure may ultimately become:

The lowest token cost per watt and the highest output per watt.

To approach that standard, operators may adopt DSX.

To reach the highest levels of performance, IREN could become the reference model.

Achieving this, however, depends on an extremely difficult-to-replicate integrated system.

Within that system, the decisive factor is not software alone.

It is not power alone.

It is not HBM memory.

It is not optical networking.

It is not any individual bottleneck currently receiving market attention.

Rather, it is the combination of all these capabilities.

The bottlenecks people focus on today are simply challenges that must be solved while building a much larger integrated capability.

Many IREN investors still believe that the company's competitive advantage comes primarily from controlling scarce resources such as power and land.

Some worry that if power shortages disappear after 2030, IREN's moat will disappear as well.

In my view, that interpretation significantly underestimates the company.

That is not what is happening.

To achieve the goal of the lowest token cost per watt and the highest output per watt, IREN has already built a three-layer system consisting of:

The energy layer
The chip layer
The infrastructure and software management layer

Its deep collaboration with NVIDIA may eventually allow participation across the broader AI stack.

Over time, IREN could gradually extend its influence toward the model layer and application layer.

Direct participation appears less likely in the near term.

However, strategic partnerships could provide indirect exposure and additional leverage opportunities.

As a result, value creation could become much larger than currently appreciated.

The market may eventually stop viewing IREN as simply an AI cloud company.

Instead, it could be redefined as:

The most important infrastructure platform company of the NVIDIA AI Factory era.

If that happens, its commercial opportunity set could expand dramatically.

First, IREN could provide DSX-standard technology licensing, facility engineering consulting, and managed operations services, generating high-margin fees and recurring operating profits.

Rather than selling compute alone, it would sell complete AI Factory templates, standards, and operating frameworks.

This could position IREN as a key partner in global Sovereign AI initiatives and place it at the center of the emerging AI infrastructure supply chain.

Second, following IREN's acquisition of Mirantis, and with the support of DSX Flex, the company could connect directly to NVIDIA's DSX Exchange communications backbone.

This would create seamless integration between the physical and software worlds.

From substation switching and microgrid dispatching at the lowest infrastructure layer, all the way up to enterprise Kubernetes deployments at the application layer, a unified control chain could emerge spanning power systems, networking, GPUs, containers, and applications.

Through extreme optimization, every watt of electricity could be converted into the maximum possible density of AI tokens.

This is infrastructure-level software optimization rather than traditional application software.

It represents a unique advantage for IREN.

Under the same electricity cost structure, IREN could produce more AI tokens than competitors.

This, in turn, could enable a unique toll-road-style economic model with substantially higher returns than peers.

Third, flexible power arbitrage.

This is already an area where IREN has extensive experience.

Once DSX Flex becomes operational, IREN's advantages as a large-scale owner and controller of power infrastructure could be maximized.

Fourth, becoming a key enabler of compute financialization and an important anchor asset within that ecosystem.

This topic deserves a separate dedicated analysis in the future.

[franklee6924x @franklee6924T]

$IREN: The Decisive Role of Data Center Technological Transformation in Enhancing Computing Power

The current market’s main impression of $IREN is that it is a company relying on pre-positioned power and land, plus vertical integration, as its core competitive advantages. In particular, the pre-occupation of land and power is the primary basis for how the market understands and values IREN. As other former Bitcoin miners shift into AI, and CRWV and NBIS successively announce plans to reach 5GW of power by 2030, IREN’s comparative advantage in this area has become less obvious.

For investors and institutions with limited knowledge, this view is understandable. But stopping at this level of understanding means missing a major investment opportunity. The AI industry has only just begun. Infrastructure development is still in a chaotic phase where each player operates based on its own understanding. The sharp surge in hype creates real challenges for both industry participants and investors in clarifying the development path and grasping the core elements.

IREN holds enormous research value because it is a long-term practitioner and deep thinker in the HPC industry. The moat it has built for itself is not physical pre-positioning but technological strength and an irreplicable scale. It never chases short-term market hype; instead, it builds its business based on its own foresight and deep understanding of the entire AI industry’s development.

Investing in the AI industry is a vast and complex undertaking. Studying IREN is an excellent entry point. We can observe and compare CRWV and NBIS, directly engage with NVIDIA and Dell, as well as Anthropic and OpenAI. We also track AMD and AVGO, and stay informed on developments at GOOG, AWS, Microsoft, and META. On top of that, sovereign AI has now emerged as a massive potential market. This multifaceted connectivity itself says a lot: IREN is not simple. Even from the limited information it has disclosed so far, a far grander and more complex future is already visible.

As mentioned earlier, the market’s mainstream view of IREN remains focused on physical pre-positioning and does not see it as a major advantage. Even after IREN announced last week that it had secured power access rights for Australia’s first 0.8GW site, some people still concluded that “this looks easy.” This shows the market’s understanding in this area is still at a very preliminary stage. Improving awareness here simply requires paying closer attention to various industry developments. In my view, this aspect is an important component of a company’s long-term moat — but only if executed properly. Pure pre-positioning advantage alone lacks sustainability.

Many people ask: once power supply constraints ease in a few years, what will IREN rely on to stand strong? This article answers exactly that question and explores what IREN’s most fundamental moat really is.

First, predicting exactly when power supply tensions will ease is extremely difficult. The current reality is that the situation is becoming increasingly tight. IREN has undoubtedly built an impressive lead here. But the key is what enduring, irreplaceable, long-term leadership actions it can take during this window of advantage — turning an effective moat into a profound, completely unreplicable long-term capability. I believe this is the fundamental criterion for judging whether a company deserves long-term investment.

The reason I have devoted significant effort to researching IREN — beyond it being a key gateway into the core AI investment circle — is that I believe this company has a profound understanding of this issue and has been putting it into deep practice. On this point, the market’s current awareness and the valuation it assigns are, in my view, zero.

IREN’s accumulated experience in data centers and the many technological transformations it has made to support AI development are the most deeply hidden aspects and the least noticed by the market. This is not because of deliberate concealment or lack of market interest, but because such a complex, systematic technological integration cannot generate matching commercial impact and attention without a landmark event to spotlight it. I believe that moment is now brewing. The “big thing” referred to in the company slogan that IREN is vigorously promoting is exactly this landmark event.

I have had many private exchanges with @brianfry01. He believes the data center industry has remained stable and conservative for many years. Their startup wanted to drive real change in the field but ran into the reality that smaller companies didn’t need it while hyperscalers prioritized stability above all. Later, their team moved to hyperscale companies and ended up with relatively little to do. After IREN acquired them, it had already implemented numerous industry transformations during the Bitcoin mining phase and achieved substantial economic benefits. I believe the era when these 20 years of accumulated efforts will truly shine is the AI industrial revolution. IREN’s modular thinking in data center design — one room adaptable to multiple chip types, allowing seamless switching between multiple users, and pushing every watt of power utilization to the extreme — is now being realized and implemented. The addition of MIRANTIS in particular will release enormous energy from this transformation.

One core purpose of IREN’s vertical integration model is to provide the necessary framework for data center technological transformation. Transformation requires rapid response; without the right model in place, change cannot happen. Therefore, when comparing competitive factors between CRWV, NBIS, and IREN, while BTM power costs are important, the technological transformation enabled by vertical integration is far more critical. This is something competitors simply cannot achieve or even imagine. Even if third-party partners deliver well, making any changes to align with superior hardware and software performance would be extremely complex and basically impossible.

IREN will prove to the world that it is the expert in AI data centers and will thereby build an irreplicable moat. This is the true core of its positioning at the center of the AI industry — and I believe this major event has already begun to unfold.

Why did NVIDIA choose only IREN to build the DSX flagship smart factory? Is it simply because the already-powered SWEETWATER site offers massive scale advantages? That is certainly an important factor, but only one of them. I believe the more important factor is technological. Let’s look at what transformative data center technologies will bring to computing power enhancement.

In the cloud computing era, power supply determined data center scale. With the arrival of NVIDIA’s GB200 NVL72 architecture and Rubin, this narrative is undergoing a qualitative shift. Once computing power enhancement enters the era of systematized GPU clusters, data centers are no longer simple warehouses of compute — they become engineering systems that enable efficient collaboration across GPU clusters of different scales. This means that, through system-level optimization, a 2GW power facility has the potential to deliver effective computing power equivalent to 4GW or even 8GW.

This is a climb up the efficiency gradient. In a small 10MW cluster, system engineering optimizations may only bring minor gains in single-machine utilization. But during expansion at 100MW, 500MW, or even 1GW scale, these system capabilities no longer produce linear returns — they deliver multiplier effects, even geometric growth.

The multiplier effect comes from multiple aspects. First is topology awareness. Communication bandwidth differences between GPUs are significant, and the larger the cluster, the greater the differences. If the scheduling system does not understand the physical topology and fails to assign tightly collaborative tasks promptly and accurately, even the most powerful GPUs will idle due to data transmission delays. Vendors with system-level scheduling can precisely “pin” compute tasks within physically neighboring NVLink domains, pushing communication efficiency close to the theoretical limit.

Second is integrated thermal management. Under prolonged high load, GPUs automatically throttle due to rising junction temperatures, causing roughly 20% computing power loss. In traditional air-cooled data centers this is unsolvable. In transformative AI data centers, however, liquid cooling infrastructure is planned together with building thermodynamics, keeping GPUs in their highest sustained performance range at all times. This not only saves energy but fundamentally reshapes computing efficiency.

Third is software-layer orchestration. In a ten-thousand-card cluster where hardware failures occur constantly, achieving super management and task allocation to ensure continuity is where the Mirantis-IREN combination will shine. Building an operating system capable of commanding hundreds of thousands of “GPU neurons,” while delivering cluster-wide efficient operation, secure isolation for users, high system utilization, and rock-solid stability — this will be the most important breakthrough in solving the industry’s key pain points.

Together, these three elements deliver a massive efficiency boost in converting power into computing power. They create a true multiplier effect and represent the best solution to today’s energy and power constraints. IREN itself also possesses additional unique advantages that further amplify this multiplier — such as the scale of single GW-level sites and the stability provided by grid power.

Why is the DSX flagship data center jointly developed by NVIDIA and IREN so significant? Because its success will provide the industry with a replicable template for AI smart factories. Leveraging IREN’s single-site capacity advantage and its rapid transformation and response capabilities in data centers, NVIDIA can test real-world performance of GPU clusters at different scales, make corresponding technical adjustments, and ultimately build a mature, market-competitive upgraded computing power system.

Why can only IREN do this? Because it can offer already-powered, high-quality single GW-level sites right now. More importantly, its years of accumulated practice in data center development and transformation, combined with its vertical integration model, give it the exact conditions needed.

Currently, although CRWV and NBIS’s software can push each GPU to its limit, they are likely to face fundamental structural obstacles in the next phase of technology — deep topology awareness, cross-node memory orchestration, large-scale failure domain autonomy, and specialized software like Mirantis that optimizes cluster quality and stability. In other words, IREN’s vertical integration model is not merely a cost advantage; more importantly, it is a rapid-response advantage. When technological transformation in data centers becomes a key driver of AI computing power growth, IREN — as a vertical asset owner with strong data center technology capabilities — can immediately implement affinity designs between building structures and server racks, rapidly improving computing efficiency.

This is something that leasing models or reliance on third-party construction simply cannot achieve. When the underlying hardware architecture of the industry undergoes major change, the old assets that were quickly leased to bypass waiting periods become obstacles to pivoting. No matter how perfect their software is, their development space is likely to be constrained by the physical hardware environment. Although AI’s commercial potential is enormous, becoming a player with true pricing power will be basically impossible for them.

At any time, in high-tech industries, it is comprehensive technical capability that forms the true foundation of differentiated competitive advantage. The edge gained from physical pre-positioning is only temporary. Only through composite capabilities can a company build the deepest, most enduring moat.

[Mark A. Cleary @MarkACleary2]

@AvivArazi Very nice 👏🏻

[Mark A. Cleary @MarkACleary2]

$NVDA is capturing about 75% of the on‑premises AI accelerator market, making it the dominant vendor for Cloud, Industrial and Enterprise (ACIE). Combined with their OEM partners (Dell, HPE, Lenovo, Cisco, Supermicro) and most importantly their DSX Flagship partner $IREN. With Mirantis on board IREN will deliver the full stack and starting in 2027 will have the greatest AI factory in the World. Sweetwater 2GW. Unmatched TCO and Token throughput. Stay focused
NVDA🤝IREN

[NVIDIA AI Infrastructure @NVIDIAAIInfra]

🏭 The AI factory is becoming the enterprise's core infrastructure.

The @RedHat AI Factory with NVIDIA brings together OpenShell, Confidential Computing, and the full AI stack so enterprises can securely run their most demanding agentic workloads at scale.

Learn more ⬇️

[NVIDIA AI Infrastructure @NVIDIAAIInfra]

🏭 The AI factory is becoming the enterprise's core infrastructure.

The @RedHat AI Factory with NVIDIA brings together OpenShell, Confidential Computing, and the full AI stack so enterprises can securely run their most demanding agentic workloads at scale.

Learn more ⬇️

[Mark A. Cleary @MarkACleary2]

@danielnewmanUV Nice work!

[Mark A. Cleary @MarkACleary2]

The $NVDA Vera CPU is a big deal

[NVIDIA AI Infrastructure @NVIDIAAIInfra]

"Agentic AI changes the role of the CPU. The CPU is now the conductor and the GPU is the orchestra". 🎼

NVIDIA Vera is the first CPU built for AI agents — purpose-built from the ground up for how AI works today.

Faster. More efficient. Ready for what's next. 🔲

#NVIDIAGTC

[Mark A. Cleary @MarkACleary2]

@needmoreinfo54 I’m glad you thought so. Thank you. Very nice of you.

[Mark A. Cleary @MarkACleary2]

This is what the focus is. $IREN $NVDA

[Mirantis @MirantisIT]

The open standard for AI infrastructure doesn't exist yet. We're building it.

Raw compute is commodifying. Hardware generations are arriving faster than architectures can absorb them. For Neoclouds and NVIDIA Cloud Partners, the real differentiator isn't the GPU anymore. It's the software running it.

That's exactly why Mirantis and @nvidia are collaborating on a validated, open-source foundation for the next generation of AI factories.

NVIDIA DSX OS and Mirantis k0rdent AI gives cloud providers:
• Full-stack automation from bare metal to AI services
• Multi-tenant GPU clouds without vendor lock-in
• A stack that evolves with NVIDIA's roadmap, from Blackwell to Vera Rubin and beyond

And it's not just on paper. @IREN_Ltd is deploying k0rdent AI with DSX OS across thousands of GPUs—a major, real-world reference implementation of open, standards-based AI infrastructure.

Every major tech wave was shaped by open standards proven at scale. We saw it with cloud. We saw it with Kubernetes. AI infrastructure is next.

We're building that standard. And the timing couldn't be better.

Read the full story: https://t.co/cQcF9fmRVY

#NVIDIA #IREN #Mirantis #AIInfrastructure #AIFactory

[Mark A. Cleary @MarkACleary2]

@bitcoinbutcher1 I’ll join you

[Mark A. Cleary @MarkACleary2]

@jukan05 NVDA makes everyone better

[Mark A. Cleary @MarkACleary2]

@bitcoinbutcher1 Yes. NVDA will guide the way to the equation answer. It will all be boiled down to: time to compute, the total cost of operation and highest token throughput. Tokens are the new currency which requires POWER(IREN)

[Mark A. Cleary @MarkACleary2]

@AvivArazi @bitcoinbutcher1 There’s a lot to talk about when it comes to IREN and NVDA. These 2 companies together are going to deliver something special.

[Mark A. Cleary @MarkACleary2]

@AvivArazi @bitcoinbutcher1 Hi, thanks for the kind words. I love the space. Looking forward to talking more about IREN and NVDA.

[Mark A. Cleary @MarkACleary2]

IREN Investor News | Financial Announcements & Updates https://t.co/0ECpst03y3

[Mark A. Cleary @MarkACleary2]

@bitcoinbutcher1 Taking care of business!

[franklee6924x @franklee6924T]

$IREN & Anthropic (Part 3): How to Maximize Mutual Benefits Through Cooperation

$IREN has already redefined itself as a platform operator. This is a massive positioning leap. Although the CEO has already made IREN’s architecture very clear to the market, without a concrete event to serve as proof, the market cannot truly feel the essential difference. Therefore, if IREN and Anthropic cooperate, it would become a very strong validation. And what would truly drive the market to change its perception is the specific structure of the cooperation itself. Only there can the market clearly see what makes IREN unique, and clearly recognize the fundamental difference between IREN and companies like Nebius and CoreWeave.

Based on the arguments from the previous two articles in this series, any cooperation between Anthropic and IREN would almost certainly involve NVIDIA as a driving force. Therefore, when analyzing what form a partnership between Anthropic and IREN might take, NVIDIA also has to be considered simultaneously. Starting from first principles, we should first look at what the maximum interests of all three parties actually are.

For NVIDIA, the goal is to ensure that the world’s most advanced AI training workloads over the next 5–10 years continue to run on NVIDIA architecture permanently, while preventing competing chips from penetrating the ecosystem. The path to achieving this is to enable Anthropic to train the strongest models within the NVIDIA ecosystem, using performance data to prove that NVIDIA architecture is irreplaceable, thereby locking in industry-wide follow-on effects. And this NVIDIA ecosystem is precisely the DSX flagship AI factory jointly developed with IREN.

For IREN, the goal is to transform physical infrastructure assets into a platform-based company deserving technology-company valuation multiples, and to complete a historic identity transition — from infrastructure supplier to sovereign compute platform — during this once-in-a-generation AI infrastructure window. The path to achieving this is to establish a three-layer vertically integrated compounding architecture, spanning land and power, data centers, GPU clusters, and software orchestration management; to secure a central position within the NVIDIA ecosystem through partnership; and to lock in top-tier frontier-model customers.

For Anthropic, the goal is to escape the strategic dilemma of “training models on a competitor’s infrastructure,” establish sovereign compute capacity that is not controlled by any competing party, and scale compute from its current bottleneck level to the magnitude required for AGI training as quickly as possible.

When the deeper interests of all three parties are placed side by side, it becomes clear that they are not only compatible, but mutually reinforcing.

The IREN of today is already very different from the company that signed with Microsoft last year. At that time, its bargaining power was limited, and it could only provide bare-metal leasing with constrained profitability. But today’s IREN is already very different from even six months ago. In fact, during the past six months, it has not signed any additional contracts. And now that IREN has positioned itself as an AI platform operator, its next contract signing would represent an entirely different tier of strategic positioning — creating a much clearer separation between IREN and companies like CoreWeave and Nebius.

So what type of cooperation is most likely?

Given that NVIDIA needs to avoid crossing the antitrust regulatory red line associated with vertically integrated chip + software + service bundled sales, the significance of IREN as an intermediary layer becomes extremely important. By facilitating cooperation between IREN and Anthropic, NVIDIA can maximize strategic benefits while simultaneously avoiding regulatory risk and fully insulating itself from potential monopoly accusations.

A possible structure for IREN-Anthropic cooperation would be to first establish a joint technical agreement, then create a priority compute-access structure for Anthropic within the Sweetwater campus to satisfy its sovereign compute requirements. Based on this foundation, the relationship could further expand into broader and larger-scale sovereign compute supply arrangements. This would become a highly complex, ultra-long-term partnership in which technology development mutually reinforces both sides while also incorporating compute-resource supply commitments.

Such a partnership would almost certainly not resemble the pure leasing contracts used by CoreWeave and Nebius. Instead, it would likely include an equity component.

The main reasons for such an inference are as follows:

1. The cooperation between both parties represents a strategic exchange that cannot be accurately quantified.

The value provided by IREN goes far beyond GPUs themselves: completely neutral infrastructure without competitive toxicity; globally scarce 2GW+ physical capacity; DSX flagship-level performance guarantees; software orchestration capabilities enabled by Mirantis; and priority access rights to a future 5GW global pipeline. Together, these elements create a form of strategic dependency that Anthropic cannot replicate elsewhere.

This type of strategic value cannot be priced solely through contracts based on GPU-hour billing. Money can measure compute power, but it cannot measure higher-dimensional value such as “irreplaceability,” “future expansion guarantees,” or “neutrality premium.” Once a cash contract becomes incapable of carrying the full value exchange, equity naturally becomes the only reasonable value carrier.

2. The IREN-NVIDIA agreement already contains a warrant structure.

This was not an accidental innovation, but rather an industry template NVIDIA intentionally seeks to establish: strategic compute agreements should contain an equity component, because purely financial contracts cannot reflect the true strategic value of infrastructure assets in the AI era.

3. Anthropic is currently seeking $50 billion in financing at a target valuation between $850 billion and $900 billion, while its current equity pricing still carries significant uncertainty.

The secondary market is already trading above a $1 trillion implied valuation, but the formal valuation has not yet been finalized through an S-1 process. This window is extremely favorable for IREN: by exchanging its current valuation — still below Anthropic’s future IPO pricing — for Anthropic exposure, IREN is effectively purchasing equity optionality in one of the world’s fastest-growing AI companies at the lowest point.

Several years from now, the value of that equity stake could far exceed all service revenue IREN might earn from the contract itself.

The reverse is also true: Anthropic could exchange equity in a currently undervalued IREN for long-term compute guarantees. Once an Anthropic contract with IREN is publicly announced, IREN’s valuation would likely be substantially re-rated upward. In effect, Anthropic would be locking in the equity price before that revaluation occurs.

In both directions, the equity exchange contains significant arbitrage potential. This means embedding equity into the relationship creates a financially positive outcome for both parties — not merely a strategic one.

4. IREN has already secured 5GW of compute capacity across the United States, Canada, and Spain, and is developing plans globally that extend well beyond 5GW.

Anthropic’s future compute demand in Europe and Asia-Pacific must begin being planned today. A simple Sweetwater leasing agreement cannot provide Anthropic with priority access rights to IREN’s future global pipeline. That future value component cannot be priced through a traditional cash contract.

Warrants become the only mechanism capable of locking in that option value.

If Anthropic holds IREN warrants, every new IREN site that comes online globally would automatically provide Anthropic with priority access rights — because as a shareholder, Anthropic’s interests would naturally align with IREN’s global expansion interests. In practical terms, Anthropic would be using a single warrant structure to acquire a long-term option on a globally distributed sovereign compute network.

Finally, IREN’s MIRANTIS software layer is also something Anthropic would likely require.

As the AI factory paradigm standard based on the IREN-NVIDIA DSX architecture becomes established, the system will inevitably evolve toward a multi-tenant model. Anthropic would likely become a technical participant and collaborative developer within that ecosystem.

From this perspective, CoreWeave’s software capabilities cannot solve the problem of layered compute-access architecture. It cannot intelligently shift compute resources to other customers during Anthropic’s training gaps, nor can it provide unified management across Sweetwater and other global sites. Meanwhile, Nebius’ software stack is essentially irrelevant to Anthropic’s needs. That said, CoreWeave still retains significant value when handling single-purpose training tasks.

This is a topic substantial enough to deserve a dedicated article of its own, and will therefore be explored in the next piece.

At this point, the speculative discussion surrounding a potential IREN-Anthropic partnership comes to an end. Now we wait to see how actual developments unfold.