How CoreWeave and Poolside Are Teaming Up in West Texas to Build the Next Generation of AI Data Centers

In the evolving landscape of artificial-intelligence infrastructure, a singular truth is emerging: access to cutting-edge silicon and massive GPU clusters is no longer enough by itself. For companies chasing the frontier of multi-trillion-parameter model training and agentic AI deployment, the bottleneck increasingly lies not just in compute, but in the seamless integration of compute + power + data center scale.

The latest chapter in this story is the collaboration between CoreWeave and Poolside, culminating in the launch of Project Horizon, a 2-gigawatt AI-campus build in West Texas.

Setting the Stage: Who’s Involved, and Why It Matters

CoreWeave (NASDAQ: CRWV) has positioned itself as “The Essential Cloud for AI™” — a company founded in 2017, publicly listed in March 2025, and aggressively building out its footprint of ultra-high-performance infrastructure. 

One of its strategic moves: in July 2025 CoreWeave struck a definitive agreement to acquire Core Scientific (NASDAQ: CORZ) in an all-stock transaction. Through that deal, CoreWeave gains grip over approximately 1.3 GW of gross power across Core Scientific’s nationwide data center footprint, plus more than 1 GW of expansion potential. 

That acquisition underlines a broader trend: AI-specialist clouds are no longer renting space and power; they’re working to own or tightly control it.

Poolside, founded in 2023, is a foundation-model company with an ambitious mission: building artificial general intelligence (AGI) and deploying enterprise-scale agents. 

According to Poolside’s blog:

“When people ask what it takes to build frontier AI … the focus is usually on the model … but that’s only half the story. The other half is infrastructure. If you don’t control your infrastructure, you don’t control your destiny—and you don’t have a shot at the frontier.” 

Simply put: if you’re chasing multi-trillion-parameter models, you need both the compute horsepower and the power infrastructure; and ideally, tight vertical integration.

Together, the CoreWeave–Poolside announcement reveals a blueprint for that integration, and the launch vehicle is Project Horizon.

The Announcement: Partnership in Three Acts

On October 15, CoreWeave issued a public press release announcing its partnership with Poolside. 

Key elements:

• CoreWeave will provide Poolside with a state-of-the-art cluster of NVIDIA GB300 NVL72 systems, totalling more than 40,000 GPUs. 
• CoreWeave will serve as the anchor tenant and operational partner for the first phase of Poolside’s Project Horizon flagship campus in West Texas. That first phase is 250 MW of gross power, with an option to expand another 500 MW beyond that. 
• Poolside will build a 2 gigawatt (GW) AI campus on 568 acres of development-ready land in West Texas, to be developed in eight phases of 250 MW each. The site sits adjacent to a major natural‐gas hub, with dual long-haul fiber routes, established water and processing infrastructure. 

As Poolside’s Co-CEO Eiso Kant put it: 

“To compete at the frontier you need to be vertically integrated from dirt to intelligence… This partnership with CoreWeave ensures immediate access to next-generation silicon, enabling us to train multi-trillion parameter models with large scale reinforcement learning. With this secured, we can now demonstrate our compounding research, engineering, and data advantages.” 

CoreWeave’s CEO Michael Intrator said: 

“Poolside’s approach to foundation models represents exactly the kind of breakthrough work the CoreWeave platform was designed to accelerate. Together, we’re building the performance layer that will define how AI research becomes real-world capability.” 

So the announcement delivers both the compute deal (40,000+ GPUs now) and the long-term infrastructure play (Project Horizon, first phase 250 MW). But why does this matter, and what are the bigger implications?

Why This Deal Resonates in the Data Center / AI Infrastructure Ecosystem

1. Vertical integration for AI at scale.
Much like hyperscalers have long since integrated from servers through networking through power, this deal typifies the shift: AI-specific clouds (CoreWeave) aligning with AI model companies (Poolside) to control the entire stack: silicon, compute cluster, data center, and power.

The Poolside blog makes it explicit: “This campus marks the next logical step … owning not just the intelligence layer, but the energy and compute foundation that supports it.” 

In other words, vertical integration isn’t just "nice to have." The firms believe it is mission-critical to remaining competitive at the frontier.

2. The power bottleneck is now the focal point.
AI training at scale is less about renting racks and more about access to stable, large-scale power at controllable cost. Project Horizon addresses that directly: behind-the-meter modular data halls, on-site generation (aero-derivative turbines, SCR systems to meet emissions), battery storage, grid-interconnect redundancy. 

By placing the campus in West Texas, next to major natural-gas infrastructure in the Permian Basin, they aim to solve for power cost, scale and reliability. 

For CoreWeave, this ties back to its earlier acquisition of Core Scientific: gaining power-footprint optionality (1.3 GW plus 1 GW+ for expansion) means they are preparing for this kind of scale. 

3. Modular build, speed to capacity, and scale alignment.
Poolside’s blog emphasises that Horizon is built for “parallelization rather than sequential construction.” 

Each 250 MW phase can come online independently; compute clusters can begin training even as subsequent modules are fabricated and delivered. This matters because the longer it takes to ramp, the more risk companies face from silicon change-cycles, energy cost swings, or competitive displacement.

For CoreWeave, that means its infrastructure-business model must accommodate not just large capacity but agile, high-throughput rollout. Their earlier claim of being the first cloud provider to deploy NVIDIA GB300 NVL72 systems in July 2025 underlines that agility. 

4. Strategic timing and standing among competitors.
CoreWeave’s move with Poolside comes at a time when major cloud or infrastructure deals are proliferating: for example, CoreWeave recently signed multi-billion-dollar deals with major customers (e.g., reportedly Meta at ~$14.2 billion) 

The Poolside deal signals that the company is not only filling third-party demand, but also anchoring long-term, vertically-integrated partner campuses. For Poolside, this gives immediate access to frontier silicon and anchor tenancy at a campus designed from the ground up for AI.

Given the broader data center and AI infrastructure race, this deal positions both firms to compete at the frontier rather than play catch-up.

Connecting the Dots with Core Scientific

It’s worth framing this partnership through the lens of CoreWeave’s earlier move to acquire Core Scientific. That acquisition was not just about expansion. It was fundamentally about control of infrastructure (power, site control, lease elimination) in order to underpin growth of AI/HPC workloads. 

In the July 2025 release, CoreWeave CEO Intrator noted:

“Verticalizing the ownership of Core Scientific’s high-performance data center infrastructure enables CoreWeave to significantly enhance operating efficiency and de-risk our future expansion, solidifying our growth trajectory. Owning this foundational layer… will enhance our performance and expertise…” 

Thus, the Poolside announcement can be seen as the operational manifestation of that strategy. Now that the infrastructure footprint via Core Scientific and internal builds is in place, CoreWeave is using it in partnership, rather than purely as landlord, to anchor next-gen AI build-outs.

The scenario also underscores the nature of the AI era: platform clouds + model companies + infrastructure ownership are no longer separate silos. CoreWeave’s footprint via Core Scientific provides optionality; its partnership with Poolside gives direction.

Project Horizon: A Deep Dive

Here are the salient details of Project Horizon, pulled from the Poolside blog and accompanying announcements:

• 2 GW AI campus in West Texas on 568 acres, developed in eight phases of 250 MW each. 
• First phase: 250 MW gross power, with an option to expand by an additional 500 MW (making 750 MW in the near-term) with CoreWeave as anchor tenant under a 15-year lease. 
• Site advantages: Adjacent to a major natural-gas hub (Permian Basin), established water/processing infrastructure, dual long-haul fiber routes allowing ultra-high-bandwidth, low-latency connectivity. 
• Modular, factory-integrated build model: Electrical, cooling, compute are pre-fabricated off-site; each module can energize and begin operation independently. This allows “continuous, compounding deployment instead of the stop-and-wait cycle that slows most projects.” 
• Behind-the-meter architecture: Onsite generation (aero-derivative turbines with SCR emissions control), battery storage for load-balancing, grid interconnect redundancy. 
• Local community engagement: Partnering with the Mitchell family (Longfellow Ranch owner) and Pecos County to deliver jobs, infrastructure investment, training pipelines, and sustainable data center operations. 
• Timeline: The GPU cluster (40,000+ GB300 NVL72 GPUs) will reportedly come online beginning December 2025 to support Poolside’s model training. 

For Data Center Frontier readers, what this all means is evident: this is one of the most ambitious AI data center and gigawatt campus builds to date.

A 2 GW campus geared specifically for AI training/agent workloads with compute, power, network and build-method engineered from the ground up? It's a different class of facility than traditional cloud-adjacent data center expansions.

Strategic Implications 

For CoreWeave:

• Anchor tenancy of Project Horizon strengthens its long-term infrastructure presence and permits operational control rather than pure leasing.
• The 40,000+ GPU cluster deal with Poolside gives an immediate revenue and compute ramp, strengthening backlog and console of AI-training clients.
• Leverages the Core Scientific acquisition to execute on large-scale power/control strategy, making the 2 GW campus more than symbolic.
• Reinforces its positioning as a cloud partner not just for third-party APis but for deep model-training, agent-scale infrastructure.

For Poolside:

• By securing 40,000+ GPUs now, and the campus long term, it creates a competitive moat:  access to silicon + compute + power at scale.
• The vertical integration mindset (“from dirt to intelligence”) signals that Poolside sees compute infrastructure as a strategic asset, not a commodity.
• The modular build and West Texas site gives it scale-advantage, cost and power-advantage, and maybe latency/network-advantage.
• If successful, the company may become a powerful model-training platform in its own right, with enterprise agent deployments backed by its infrastructure footprint.

For the industry:

• The deal underscores how AI infrastructure is evolving from “rent large clusters” to “build multi-GW campuses anchored by model-companies.”
• Electricity and power chain are rising quickly to the top of the agenda. As you may have noticed, data center announcements now increasingly emphasize factors of onsite generation, behind-the-meter provisioning, fuel sourcing, and dual fiber routes.
• The modular, factory-prefab approach to data centers is gaining traction.
• The “parallelize and roll-on capacity” model may reduce time-to-capacity, which in AI is a key metric.
• The geography (West Texas, natural-gas hub, large land parcel) may become a blueprint for future mega-scale AI campuses.

What to Watch Next

• The actual first capacity turnover: When the 40,000+ GPU cluster goes live (expect December 2025) and how it performs (latency, throughput, model-training results).
• Execution of the first 250 MW phase of Project Horizon: Ground-breaking, construction schedule, power commissioning, module roll-ons.
• How the traditional leasing/colocation model responds: Will operators seek similar integrations with model-companies?
• Cost dynamics: Power cost, PUE, silicon-utilization, amortization of modules. These will determine whether multi-GW AI campuses are economically sustainable.
• Ecosystem developments: Will CoreWeave and Poolside form an ecosystem of other model houses, or is this more of a bespoke pair? Will competitors respond with similar commitments?

Conclusion: Infrastructure Is Destiny 

The CoreWeave–Poolside partnership, anchored around Project Horizon, represents a significant inflection in the data center/AI infrastructure narrative. Rather than incremental expansion of rack count, this is a full-scale orchestration of compute, power and build methodology tailored for AI’s next phase. 

Through the lenses of CoreWeave’s infrastructure ambitions (as bolstered by the Core Scientific acquisition) and Poolside’s model-training aspirations, the deal illustrates how the AI frontier is being defined not just by models and chips, but by underlying infrastructure control.

In many ways, as Poolside observed in its blog, you either control your infrastructure or you don’t control your destiny. For both CoreWeave and Poolside, Project Horizon embodies that proposition.

As the rollout begins in December and builds into 2027 and beyond, the outcome may become a benchmark for how AI campuses of the future are built.

At Data Center Frontier, we talk the industry talk and walk the industry walk. In that spirit, DCF Staff members may occasionally use AI tools to assist with content. Elements of this article were created with help from OpenAI's GPT5.

Keep pace with the fast-moving world of data centers and cloud computing by connecting with Data Center Frontier on LinkedIn, following us on X/Twitter and Facebook, as well as on BlueSky, and signing up for our weekly newsletters using the form below.