US Tech Wake-Up: Wall Street's AI Power Play & the Rise of Sovereign AI

Nvidia just brought in Blackstone, Goldman Sachs, and KKR to mobilize over half a trillion dollars for AI infrastructure. In episode 78, we talked about venture capital funneling into just a few AI titans. Well, that was just the start. The money isn't just coming from VCs anymore — it's coming from the absolute biggest players on Wall Street. And that's just the lead. The entire ecosystem is shifting. The market for what's called "Sovereign AI"—where nations build their own AI infrastructure—is now projected to explode from about twenty-five billion dollars this year to over three hundred billion by 2040. Governments are starting to treat compute like a strategic national asset. To power all this, venture capital investment in U.S.

nuclear energy startups has already topped four-point-five billion dollars in 2026. That's on track to beat last year's record, with figures like Sam Altman saying there's no way to meet AI's energy needs without a breakthrough. Meanwhile, the U.S. Commerce Department has started an investigation into how Chinese AI firms are accessing advanced Nvidia processors through overseas data centers, which could complicate everything. And because Congress is stalled on AI legislation, the private sector is just building its own regulatory frameworks. The Secure Technology Alliance just launched a new forum to do exactly that. You can see the physical result of all this spending in the backlogs of companies like Jacobs Solutions.

They're an engineering firm whose project backlog just hit a record twenty-nine billion dollars, with AI-related projects suddenly accounting for eleven percent of their revenue. Okay, so let's go back to that Nvidia deal. Because the half-a-trillion-dollar figure isn't even the real story. The real story is how that money is being deployed, and what it says about the new reality of AI. Nvidia's CEO, Jensen Huang, just reframed the entire conversation. He said, and this is the key, that we have moved from an era where companies bought chips for specific projects... to an era where "AI factories can be financed as productive infrastructure." Think about that. Not as a cost, but as an asset. A revenue-generating asset.

He said this is the first time that computer chips have become an INVESTABLE asset class. Here’s what that means. For decades, a server was an expense. You bought it, you used it for a few years, you wrote it off. It was like office furniture. But Huang is now arguing that Nvidia's hardware is different. It's "revenue-generating," it's "long-lived," and it's "fungible"—meaning it can be repurposed for different AI tasks. So Wall Street firms like BlackRock and Goldman Sachs aren't just helping companies buy more chips. They are creating financial instruments to fund the construction of massive, centralized AI compute facilities—the "factories"—that will be leased out, just like a power plant or a toll road.

Nvidia is even providing some financial backing to make lenders comfortable. This is the only way to meet the demand. The big cloud providers—the hyperscalers—are already planning to spend somewhere around seven hundred and forty billion dollars on AI capital expenditures. This new financing model is designed to pour fuel on that fire by bringing in long-term institutional capital that would normally fund bridges and airports. We're no longer just buying computers. We're financing the engines of the entire next economy.