The Stargate initiative, once heralded as a transformative $500 billion AI infrastructure moonshot, is stumbling under the weight of its own ambition. Originally backed by OpenAI, SoftBank, Oracle, and MGX (a UAE sovereign-linked fund), the project was intended to revolutionize global AI capacity with a network of exascale data centers stretching across the United States, beginning with a massive 1.2-gigawatt facility in Texas. That vision is now being reined in. Reports from the Wall Street Journal and Reuters confirm that Stargate is being dramatically scaled down, with current focus redirected to a comparatively modest data center in Ohio set for completion by year’s end.
This reversal in scale follows internal disputes, strategic hesitations, and a mounting chorus of skepticism from the tech elite. Elon Musk has voiced pointed concerns, suggesting the project’s backers “simply don’t have the money” to deliver on such an audacious infrastructure promise. Behind the scenes, tensions between Sam Altman of OpenAI and Masayoshi Son of SoftBank—both driving personalities with diverging philosophies—have led to friction, complicating funding flows and undermining the project’s cohesion. The fallout has not gone unnoticed in the stock market. Nvidia and AMD, both seen as crucial hardware suppliers for Stargate’s computational backbone, have faced share price dips as investors recalibrated expectations. Nvidia, which had been positioned as a primary beneficiary of Stargate’s future demand, saw its shares trade down to around $166.93, off nearly 2.6% intraday. AMD, despite Oracle’s initial intentions to integrate AMD’s Instinct MI350 chips into Stargate’s infrastructure, slipped over 3%, trading at $152.27.
Analysts had until recently cited Stargate as a powerful long-term growth lever for both chipmakers. Nvidia’s Blackwell GPU line was expected to be central to the venture, while AMD was seen as a cost-effective challenger offering compelling hardware options. However, AMD continues to struggle with software maturity, particularly when measured against Nvidia’s dominant CUDA ecosystem. Jefferies recently downgraded AMD, citing the MI300X’s performance limitations, supply-chain volatility, and a cautious view of enterprise data center demand.
Despite this turbulence, the Stargate initiative is not entirely derailed. Both WSJ and Reuters emphasize that the project’s refocus on Ohio is not a retreat but a strategic regrouping. Smaller builds in additional states are reportedly in the works, and ongoing site assessments are being conducted “with urgency.” This phased approach—scaling down to scale up—could serve as a proof-of-concept. If the Ohio build is executed successfully and demonstrates technical value and political stability, additional investment may follow. That, in turn, could revitalize the broader vision, albeit in a more realistic, modular form.
Still, the path ahead is fraught. Execution risk looms large. The fragmented nature of Stargate’s consortium—corporate, governmental, and private—complicates decision-making. Public skepticism, amplified by high-profile voices like Musk and the optics of excessive ambition, could chill future capital inflows. Investors hoping for a quick rebound in Nvidia or AMD tied to Stargate’s infrastructure demand may need to temper their expectations or pivot focus to other AI catalysts.
Ultimately, Stargate remains alive but is no longer the juggernaut it once aspired to be. Its scaled-down form reflects the sobering limits of even the most well-funded techno-utopian projects when ambition outpaces coordination. For Nvidia and AMD, the dream of a trillion-dollar AI data infrastructure race may still live on—but not without detours, delays, and the constant need to prove, rather than promise, what’s possible.