© DjelicS / E+ via Getty Images
Shares of NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) are down 5% Monday morning to $197, leading a broad pullback across AI-hardware names after The Wall Street Journal reported that NVIDIA is in talks to guarantee up to $250 billion in financing for OpenAI’s data-center buildout. The report has reignited “circular financing” concerns tied to the AI capex cycle.
The selling extends beyond NVIDIA stock. Advanced Micro Devices (NASDAQ:AMD) shares are down 8% to $479, Intel (NASDAQ:INTC) shares are down 4% to $89, and Dell Technologies (NYSE:DELL) stock is down 4% to $421. Each of these moves lands inside massive year-to-date (YTD) rallies, so today looks more like a rethink than a break in the AI trade.
A $250 Billion Financing Report Sparks the Selloff
The trigger is the WSJ scoop that NVIDIA would effectively backstop OpenAI’s compute buildout at a $250 billion scale. Investors are worried this looks like vendor financing on a historic level, with NVIDIA funding a customer that then buys NVIDIA GPUs and builds data centers packed with them.
The optics matter because AI capex has already ballooned. NVIDIA’s Q1 FY2027 data center revenue hit $75.25 billion, up 92% year over year (YoY), with CEO Jensen Huang describing “the buildout of AI factories, the largest infrastructure expansion in human history.” That framing sounds bullish when demand is organic and circular when the seller is helping to fund the buyer.
Sympathy Selling, With One Exception
AMD stock is taking the hardest hit because it competes directly for AI-accelerator dollars, and any wobble in the buildout thesis pressures the MI450 growth narrative that helped drive AMD shares up 124% YTD. Dell shares are sliding on the same worry from the systems side, since Dell books AI-server orders tied to data-center capex, and management guided AI-server revenue of $60 billion for FY2027.
Intel is the odd one out. INTC shares are mostly giving back part of Friday’s post-earnings pop, with today’s move driven by profit-taking rather than the OpenAI story. The company’s Q2 2026 revenue came in at $16.13 billion, up 25.4% YoY, with Data Center and AI up 59% YoY to $6.26 billion, and Intel was named host CPU for NVIDIA’s DGX Rubin NVL8 systems. The pullback in Intel shares appears to be a function of profit-taking and broader chip softness.
Capex Fatigue Sets In Across the AI Complex
Oracle (NYSE:ORCL) shares are up 4% Monday but have plunged recently on OpenAI and data-center exposure, down 20% over the past month. Alphabet (NASDAQ:GOOGL) shares are up 3% today after the company again raised its capital-expenditure forecast, and Microsoft (NASDAQ:MSFT), the other major OpenAI backer, sits at the center of the same debate.
Traders holding the iShares Semiconductor ETF (NASDAQ:SOXX) are feeling this directly, since the fund holds NVIDIA, AMD, and Intel among its top names. The SOXX ETF holds a narrow, concentrated basket, and days like today are a reminder that a sector-specific fund can move sharply on a single storyline. Hence, investors should size their positions accordingly.
Bull Case, Bear Case, and What to Watch
The bull case is intact on the numbers. NVIDIA still guides $91 billion in Q2 FY2027 revenue, AMD’s data center segment grew 57% YoY last quarter, and Dell booked $24.4 billion of AI orders in a single quarter. Demand is real, and the cash is showing up on the income statement.
The bear case centers on structure. If more of the AI buildout depends on the seller financing the buyer, the market has a fair reason to question how much of the reported revenue is truly independent. The prediction markets are already reflecting that mood, with Polymarket pricing a 98% probability that NVDA closes down today.
Investors may want to keep their position sizes modest here and watch for whether NVIDIA confirms or clarifies the WSJ report, whether AMD’s early-August earnings reinforce the AI-accelerator demand narrative, and whether Oracle’s, Alphabet’s, and Microsoft’s commentary on the next round of calls addresses capex return on investment. The next data point could reset the tone fast.
Contact [email protected] for any questions or corrections.
