Nvidia’s Summer of Giants: From SpaceX’s 10-Gigawatt Bet to a Skeptic’s $250 Billion Warning

Elon Musk has a way of resetting the conversation. When SpaceX announced it would anchor its future AI computing capacity exclusively on Nvidia’s Vera Rubin architecture — scaling from 2 gigawatts of installed power by the end of 2026 to 10 gigawatts by 2027 — the message was unmistakable: this is infrastructure on the scale of a utility, not a product launch. For a company that began as a graphics card maker, the endorsement marks a remarkable evolution into the foundation layer of an entire industry.

The timing is no accident. Vera Rubin entered full production in late May, fabricated on TSMC’s 3-nanometer process with HBM4 memory stacks. By mid-June, a Quanta Computer executive confirmed the first production units of the Rubin GPU — internally designated R200/VR200 — would reach customers by August. The sequence tells its own story: manufacturing readiness, then deliveries, then a marquee customer publicly committing its entire infrastructure roadmap to the platform. It is a carefully orchestrated rollout, and it raises an obvious question for investors: how many more SpaceX-style moments are still to come?

The Deal Machine Keeps Running

The SpaceX announcement is hardly an isolated event. Late July brought a blockbuster: SK Group and Nvidia unveiled a collaboration valued at more than $500 billion. SK Telecom is slated to build a “Vera Rubin DSX AI Factory” with two gigawatts of capacity, while SK hynix confirmed a long-term partnership to develop the next generation of HBM memory. A day earlier, Nvidia had announced a long-term partnership with Safe Superintelligence Inc. to accelerate the AI startup’s growth. Then, in early August, Nvidia joined forces with 36 other organizations to launch the “Open Secure AI Alliance,” focused on open technologies for securing software and AI agents.

For bulls, this deal flow is the story. It suggests demand for Nvidia’s technology extends far beyond chip shipments into the very architecture of AI infrastructure. Skeptics would need to explain away a string of multi-billion-dollar commitments to argue that the cycle is cooling.

The Bear Case Has a Name

That skepticism has a prominent voice. Michael Burry, the investor known for betting against the housing bubble, has warned of a $250 billion demand risk hanging over the sector. His caution cuts against a narrative that has pointed almost exclusively upward for months. To be sure, the consensus remains firmly bullish — 36 analysts currently rate the stock a buy — but consensus opinions have a way of being wrong at precisely the wrong moment.

The insider activity adds a subtle counterpoint. Over the past 90 days, Nvidia insiders have registered 13 transactions, all of them sales, totaling roughly $767.2 million. Not a single insider bought shares during that period. In late June, board member Mark A. Stevens sold 885,000 shares for about $186 million, according to a mandatory SEC filing. Such moves are routine for executives diversifying their wealth, but they sit awkwardly alongside the euphoria narrative.

Should investors sell immediately? Or is it worth buying Nvidia?

A Stock That Has Already Priced in a Lot

The market, for now, is siding with the optimists. The stock closed Thursday at €189.82, just 6.26 percent below its 52-week high of €202.50. Over the past seven trading days alone, the shares have gained 9.02 percent — a pace that suggests investors are already celebrating the recent partnership announcements. The stock trades comfortably above its short- and medium-term moving averages, signaling intact upward momentum but leaving little room for disappointment.

The broader picture is equally striking. In early August, Nvidia reclaimed its position as the world’s most valuable publicly traded company after a pullback in Apple’s shares, reaching a market capitalization of $4.86 trillion — roughly €4,444.49 billion. It is a symbolic title, but one that underscores how tightly the company’s fortunes are now woven into the broader AI narrative.

Wall Street has taken notice. JPMorgan and Citi both raised their price targets on Tuesday, citing strong cloud results from customers Amazon and Microsoft as a green light for continued AI infrastructure spending. Other firms have followed suit in recent days. The shared logic: as long as hyperscalers keep pouring money into data centers, Nvidia remains the central beneficiary.

The Moment of Truth Arrives August 26

All of this sets up a pivotal date. On August 26, Nvidia will report second-quarter results for fiscal year 2027, covering the period through July 26. The market expects revenue of around $91 billion, according to media reports. That number will be the real test: it must translate the recent flood of deals into hard growth figures. If it falls short, Burry’s warning could suddenly carry more weight than the market currently assigns it.

Beyond the earnings call, the calendar offers another marker. From October 20 to 22, Nvidia will host GTC Europe in Berlin, where it plans to present its full technology roadmap. Coming after July’s SIGGRAPH conference — where the company showcased new techniques for neural rendering and “Physical AI” world models for autonomous systems and robotics — Berlin should demonstrate just how far the company’s ambitions have moved beyond graphics chips.

The central tension is now clear. The structural arguments are compelling: the alliances with SK Group, SSI, SpaceX, and a broad swath of industry show Nvidia systematically building out its ecosystem, not merely filling its order book. Yet the insider sales and Burry’s warning serve as reminders that the valuation leaves little buffer for disappointment. August 26 will reveal which of the two narratives is closer to reality — and whether the summer of giant deals can survive contact with an actual earnings report.