Key Highlights
- On November 16, 2026, Oracle (NYSE:ORCL) slid 4.99%, Dell Technologies lost 6.96%, Astera Labs 8.79%, DigitalOcean 8.62% and GE Vernova 4.9%.
- An industry research firm is reported to have lowered its 2028 forecast for global AI accelerator spending by 28%, to about USD 758 billion, and its 2030 forecast by about 40%, to USD 923 billion.
- Although the figures stem from summaries of a subscriber-only model and lack public confirmation, they still set off a broad sell-off in AI-linked shares.
- Technology was the weakest S&P 500 sector with a 3.24% drop, and a 10-year Treasury Yield holding near 5.12% weighed further on capital-intensive businesses.
1. Oracle (NYSE:ORCL)
Oracle (NYSE:ORCL) dropped 4.99% on November 16, as it has the heaviest Balance Sheet exposure of the five. A very large build-out of cloud data centers for AI customers has been committed to by the company, partly financed with debt, so any suggestion of softer Demand prompts questions about what that spending will return. In the reported forecast cut, 2028 accelerator spending lands at about USD 758 billion, 2029 comes in about 41% lower at USD 796 billion, and 2030 stands at USD 923 billion. Still, 2030 spending would be more than double the 2026 estimate of about USD 392 billion, which makes the slope of growth the issue and not its direction. Interest rates sharpen the debate. A 10-year yield near 5.12% and a December Federal Reserve hike priced as highly likely mean that borrowing to build capacity costs more than a year ago. Some factors offset this. Anthropic plans to price its IPO on November 18 and has disclosed about USD 900.9 billion of purchase and Lease commitments, almost all for compute, evidence that big buyers continue to lock in capacity years in advance. To see whether contracted demand covers the build-out, investors will follow Oracle's financing activity and cloud backlog.
2. Dell Technologies (NYSE:DELL)
As one of the biggest suppliers of AI servers, Dell Technologies (NYSE:DELL) fell 6.96%, harder than Oracle. Server makers stand squarely in the path of accelerator spending, since every rack they ship holds the processors, memory and networking that the forecast revision covers. The research firm is reported to have given three reasons for the cut: slower Revenue growth at frontier AI labs, tighter financing for data centers lacking long-term offtake agreements from large cloud providers, and delays affecting about 5.8 gigawatts of US Data Center capacity. For Dell the second reason matters most, because smaller cloud and enterprise customers rely on financing to place large orders. A policy layer comes from the new Commerce Department rule. Licences are required for advanced AI chips shipped to data centers in five countries, Singapore and Malaysia among them, a region that has grown important for data center construction, so server orders linked to those sites may be delayed. The main test is Dell's next update on AI server orders and backlog. Because the two trade on the same demand signal, a strong Nvidia report after the close on November 17 would probably help Dell, while caution on 2027 orders would deepen Monday's losses.
3. Astera Labs (NASDAQ:ALAB)
The steepest decline of the five, 8.79%, belonged to Astera Labs (NASDAQ:ALAB). Its connectivity chips link processors inside AI systems, a fast-growing niche that ranked among the strongest performers of the AI trade, and that success left it exposed. Outer-year estimates move stocks priced on years of rapid growth the most, and the reported revision lowered expected growth to about 21% in 2028 and about 5% in 2029, from 37% and 28% before. Across the wider Supply chain, the decline had company: Arm fell 8.12%, Lumentum 8.48%, Ciena 8.19% and Sandisk 8.58%, while a three times leveraged semiconductor fund lost 16.39% and its inverse counterpart gained 16.54%, a sign that crowded positioning was unwinding fast. Astera's products ride along with accelerator shipments, so its revenue tends to track the build-out of large clusters closely. Nvidia's remarks on rack-scale systems, and on the mix between full systems and individual chips, are therefore especially pertinent. Should Nvidia signal that hyperscale customers have not paused orders, connectivity suppliers could bounce sharply from oversold levels. Should it hint at delays, Astera's rich valuation gives it the most room to fall among these five.
4. DigitalOcean (NYSE:DOCN)
DigitalOcean (NYSE:DOCN) gave up 8.62%. Having pushed into AI computing by renting capacity to developers and smaller companies, the cloud provider is a second-tier beneficiary of the same theme that lifted larger operators. One risk in the forecast revision hits smaller cloud providers in particular: tighter financing for data centers without long-term contracts from large cloud customers. Rising rates also reach them more directly, since expansion often relies on borrowing or Equity raises. Institutional appetite was already cooling before this month's sell-off, based on quarterly holdings filings published on November 16 and reviewed since. Of nearly 6,400 filings analysed, at least 138 showed funds fully exiting data center stocks in the third quarter against 119 that opened positions, and the count of holders across eight large data center names slipped to 3,392 from 3,411. As those findings circulated, CoreWeave edged down about 2% on November 17 before the open. What DigitalOcean must show is that demand from its developer customer base holds if AI spending growth slows at the top of the market. It cannot match the largest clouds on pricing given its smaller scale, so it leans on service and simplicity. Steady usage growth would set it apart from the broader sell-off.
5. GE Vernova (NYSE:GEV)
The sell-off reached power equipment, as GE Vernova (NYSE:GEV) lost 4.9%. Gas turbines and grid equipment that data centers need for electricity are its products, a part of the AI build-out that investors had considered sturdier than chips. Holdings filings reveal that some funds were already trimming power exposure, with 212 funds exiting Utility and power stocks in the third quarter and only 96 opening positions. Other power suppliers offer a useful contrast. Powell Industries, which makes switchgear and power control rooms, reported after the close on November 16 that fiscal fourth quarter orders climbed 162% to USD 710 million, including a data center order of about USD 85 million, and that Backlog hit a record USD 2.7 billion. Contracted demand for electrical equipment therefore looks strong even as forecasts for accelerator spending fall. Delays to about 5.8 gigawatts of US data center capacity were also cited in the reported forecast, and these might stretch delivery schedules for power equipment without cancelling it. Long-cycle orders give GE Vernova more visibility than chip suppliers enjoy, which explains why it fell by less than Astera or DigitalOcean. Whether order commentary from utilities and data center developers stays firm is the next test.
Where Things Stand
GE Vernova and Oracle fell the least among the top 5 AI infrastructure stocks hit by the forecast cut and have the most contracted demand behind them, whereas Astera Labs and DigitalOcean carry the greatest valuation and financing risk. Dell falls in between as a direct read on server orders. Whether AI spending is slowing as the revised model suggests will be tested next by Nvidia's results after the close on November 17 and by the pricing of Anthropic's IPO on November 18, respectively.






Please wait processing your request...