AI's New Oil: Infrastructure Deals Erupt Past $5B in a Week

The Great Compute Land Grab Is Here

Forget the shiny apps. The real money in AI is going somewhere else entirely. It's going to the plumbers. A gusher of capital just hit the AI infrastructure sector, with deals so big they’re changing how everyone thinks about the raw power behind the bots. And this isn't your typical Sand Hill Road venture cash. We're talking about a strategic flood from energy giants and institutional players who now see AI compute for what it is: a utility. An asset as essential as oil.

Exhibit A: Together AI. The San Francisco platform for open-source models just closed a monster $800 million Series C on July 1, 2026. And get this—the round was led by Aramco Ventures, the VC arm of Saudi Arabia's state oil company. That one move more than doubled Together AI’s valuation to a staggering $8.3 billion, up from $3.3 billion just 16 months ago. The investor list reads like a who's-who of tech and finance—Nvidia, Vista Equity Partners, General Catalyst. You name it. When an oil titan leads one of the biggest AI infrastructure deals ever, that’s not just a check. It’s a statement. Vipul Ved Prakash, Together AI's co-founder and CEO, put it bluntly: “Intelligence is becoming a foundational resource… Our mission is to ensure that intelligence is abundant, not expensive.”

Their momentum is just insane. Annualized bookings blew past $1.15 billion last quarter. Together AI is making a simple, powerful bet: companies are sick of the sky-high costs and vendor lock-in from proprietary giants like OpenAI and Anthropic. So they'll run to open-source models instead. By building the specialized cloud to run them, Together AI claims it can cut the cost of actually using these models—the inference—by a wild 60x.

Valuations Triple as Power Becomes the Prize

Just as the ink was drying on the Together AI deal, whispers of something even bigger started. Much bigger. Crusoe, an AI data center developer, is supposedly talking about a new round. The target? A cool $3 billion. That would rocket its valuation to somewhere around $30 billion—a 3x jump from its $10 billion value in October 2025. This isn't just growth; it's an explosion, fueled by an insatiable hunger for raw power and data center space.

Crusoe's pivot has been nothing short of spectacular. It started back in 2018 with a clever idea: use flared natural gas to power crypto mining. Now? It's all about building the factories for the AI age. The company already has contracts for almost 5 gigawatts of computing power locked in, with a pipeline for more than 40 gigawatts. And who's buying? The biggest names in the business. We're hearing Meta Platforms, Oracle, Microsoft, and even OpenAI are customers. In the great AI gold rush, Crusoe's trick of turning wasted energy into AI-ready compute has made it one of the hottest tickets in town.

And it’s not just venture-backed darlings, either. Look at Switch. The data center operator is said to be chasing $2 billion in a private round led by none other than Andreessen Horowitz. A deal like that could value the Las Vegas company at nearly $50 billion, debt included. That's a mind-bending four-fold increase from its $11 billion take-private valuation in 2022. When a VC kingmaker like Andreessen Horowitz cuts a huge check for a glorified data center landlord, it’s a massive tell. The smart money believes the real, lasting value isn't in the AI models. It's in the picks and shovels—or, really, the power and cooling.

Why AI Infrastructure is Becoming a Utility Asset Class

So what's going on here? This isn't just another hot market. This is a fundamental reclassification of AI infrastructure itself. It’s no longer being treated like a speculative tech startup. It's being treated like a utility. Think power grids, pipelines, shipping ports—that essential. The investors piling in aren't your typical VCs chasing a quick flip. They're sovereign wealth funds, energy giants, and huge asset managers. They’re buying long-term, strategic capacity in the single most important resource of the new economy.

Here’s why the shift is happening:

• Scarcity of Power: Forget chips. The real bottleneck for AI is now electricity—massive, reliable sources of it. Any company that can lock down land and gigawatts of power has a license to print money. Just look at Crusoe's energy-first playbook.
• Standardization of Compute: All the AI models may be different, but what they consume is the same: GPU-hours. That's fast becoming a standardized unit, a commodity you can buy, sell, and finance. It’s only a matter of time before we see compute futures trading on an exchange.
• Sovereign Interest: This has become a national security issue. Seriously. Countries now see domestic AI compute capacity as vital for their economic future, which explains why state-linked funds like Aramco Ventures are jumping in with both feet.

The scale of this investment is just breathtaking. Morgan Stanley Research projects nearly $3 trillion will pour into AI infrastructure globally by 2028. Three trillion. This isn't just about building a few more data centers. It’s financing a build-out on the scale of an industrial revolution. As one analyst at Global Data Center Hub wrote, Q2 2026 was the moment “capital markets turned compute into a rated, tradable credit product.” It feels a lot like the early days of the AI chip war—the companies laying the foundation are setting themselves up to win big, and for a long time.

Right now, the money spigot is wide open. The race isn’t just about who can build the smartest model. It’s about who owns the plumbing. Who controls the physical grid that brings AI to life. But here's the multi-billion dollar question: how long can this spending spree last before investors start asking for their money back?