Artificial Intelligencer-From SpaceX to Nvidia, the deals showing AI runs on capital

By Krystal Hu

Feb 4 (Reuters) - (Artificial Intelligencer is published every Wednesday. Think your friend or colleague should know about us? Forward this newsletter to them. They can also subscribe here or email me to share any thoughts.)

Jensen Huang didn’t look like someone who’d been crisscrossing the globe for two straight weeks when he surprised a group of tech executives by walking into the reception of Cisco’s CSCO.O AI Summit in San Francisco on Tuesday evening.

Wearing his signature leather jacket and all-black uniform, Jensen asked for a glass of white wine and was quickly surrounded. Executives leaned in with questions, collaboration pitches, and, inevitably, selfie requests. Intel INTC.O CEO Lip-Bu Tan was no exception, snapping a photo with him amid the buzz.

By the time Jensen took the stage, wine glass in hand — he emphasized it was his fourth—it was clear this wouldn’t be a tightly scripted fireside chat. Sitting next to Cisco CEO Chuck Robbins, Jensen drifted into a philosophical mode that resisted neat Q&A framing. At one point, after Jensen asked, “Do you know what I’m saying?” Chuck shot back, half-joking, “Do you know what you’re saying?”

In this unscripted state, he was funny, candid and fond of analogies on the biggest topics in AI: how enterprises are using them.

He compared it to parenting and gardening, pushing back on the idea of the endless pursuit of ROI. “You let a thousand flowers bloom,” he said. “Later, you curate. I haven’t started curating yet—but I encourage everybody to try.”

He also waved off stock-market anxiety around software companies, arguing that software is like “calculators”—tools you don’t need to reinvent just because you now have AI. AI agents, he added, are designed to use tools, not replace them.

Perhaps more striking was his skepticism about putting everything in the cloud, even though hyperscalers are Nvidia’s NVDA.O largest customers.

“I am not secure about putting all of Nvidia’s conversations in the cloud, which is the reason why we build it locally,” he said. “I’m just not confident to share that conversation—because my conversations are the most valuable asset to me.”

How many tech leaders will follow Jensen’s approach to adopting AI inside their own companies remains an open question. In this week’s issue, we dig into the billion-dollar deals racing to meet insatiable AI demand—and why the stock market is punishing software companies along the way. Scroll on.

OUR LATEST REPORTING IN TECH AND AI:

• Exclusive-Amazon plans to use AI to speed up TV and film production

• Exclusive-OpenAI is unsatisfied with some Nvidia chips and looking for alternatives

• Exclusive-Despite new curbs, Grok produces sexualized images without consent

• As software stocks slump, investors debate AI's existential threat

• AI voice startup ElevenLabs secures $11 billion valuation in latest funding round

FROM SPACEX TO NVIDIA, AI RUNS ON CAPITAL

Old and new partnerships in AI this week point to a familiar reality: for all the talk of future ambition and breakthrough technology, this race is still very much about capital.

Take the merger of SpaceX and xAI. The news, first broken by my colleagues and confirmed by the company this week, creates what Musk has framed as a trillion-dollar entity. At its core, the move consolidates Musk’s private empire and bundles assets to accelerate xAI’s ability to raise money by capitalizing on investors’ enthusiasm for SpaceX’s cash-generating business, Starlink.

This kind of bundling is not new for Musk. SpaceX has long been considered the crown jewel of his portfolio, and its gravitational pull has helped make less attractive deals palatable to investors in the past as a way to build relationships and get allocations—including the Twitter (now X) buyout, sources previously told me.

xAI, even after merging with X, is still burning cash and carrying significant acquisition-related debt. Under the new structure, investors are expected to receive SpaceX shares. Not every SpaceX investor I spoke with welcomed the development, though most conceded this is a familiar risk across Musk’s companies: limited governance and a heavy reliance on trust in his instincts.

The pitch now, according to investors, is that the conglomerate remains on track to go public in the coming months—offering a potential payoff as retail investor interest surges around owning a piece of both space and AI.

There are, of course, more futuristic angles to the deal. Space-based data centers and “physical AI” are among the most discussed. But experts caution that practical challenges—from transporting hardware to maintaining systems in orbit—could take years to solve. Still, Musk has a track record of selling moonshot ideas, securing funding first, and working through the details later.

Another capital raise unfolding largely in public view is OpenAI’s ongoing funding round, valued at more than $800 billion. At that scale, anchoring the round with key partners is critical—and CEO Sam Altman appears to have done just that by locking in support from Amazon AMZN.O, Nvidia and SoftBank 9984.T.

SoftBank, with over $40 billion invested, is now one of OpenAI’s largest shareholders. Amazon could emerge as a strategic counterweight to Microsoft MSFT.O in OpenAI’s ecosystem. Nvidia’s role is particularly notable: while it had previously committed up to $100 billion in investment—terms that have taken longer than expected to finalize—it is now preparing to invest roughly $20 billion in this round.

For Nvidia, the strategy is clear: lock in one of its largest customers by becoming an investor. But the relationship is evolving. OpenAI, wary of being tied too closely to any single vendor, is actively seeking alternatives. While its current compute stack still relies entirely on Nvidia GPUs, we’ve reported that the company is exploring other options for inference workloads—especially those requiring ultra-fast response times to meet surging demand.

In today’s AI race, the technology may be groundbreaking—but the real story remains the capital behind it, and the strings that come attached.

CHART OF THE WEEK:

Investors are rewarding the makers of artificial intelligence hardware, sending semiconductor stocks soaring roughly 65% since early 2025, while punishing software companies on fears the technology will make their products obsolete. Software and services stocks fell 8% during that period, making them one of the weakest-performing parts of the index. The AI boom is favoring companies that sell the picks and shovels, while many software firms are being re-rated amid fears their products are easier to disrupt.

That pressure intensified this week as investors increasingly worry that new offerings from foundational model companies aren’t just enhancing software but replacing it, particularly in professional services and knowledge-work tools. Investors and tech executives told me that while the selloff looks overdone, software companies now have to prove they are systems of record, not just productivity layers. To defend their valuations, software companies are embedding AI deeper into workflows, locking in proprietary data, and showing they can evolve beyond feature-based tools.

(Reporting by Krystal Hu; Editing by Lisa Shumaker)

((krystal.hu@thomsonreuters.com, +1 917-691-1815))