Key Highlights

  • In a model shared with subscribers over the weekend of November 15, 2026, an industry research firm reportedly lowered its 2028 global AI accelerator spending forecast by 28%, to about USD 758 billion.
  • Reportedly, the 2027 forecast was trimmed 19% to USD 627 billion, 2029 was reduced 41% to USD 796 billion and 2030 about 40% to USD 923 billion, for a combined cut of roughly USD 1.6 trillion.
  • Summaries of a subscriber-only model are the channel through which the figures circulated, and the firm has not publicly confirmed them.
  • On Monday, November 16, VanEck Semiconductor ETF (NASDAQ:SMH) lost 5.11% as the report and new US export rules pressed on chip stocks around the world.

Contents of the Revised Outlook

Money spent by cloud providers, AI labs and other buyers on the processors that train and run large AI models is what accelerator spending means. Reportedly, the revised model leaves the 2026 estimate of about USD 392 billion largely as it was but lowers the path sharply from 2027 on. Growth for 2027 is still put at about 60%, yet expected growth slows to about 21% in 2028, 5% in 2029 and 16% in 2030, compared with earlier figures of 37% for 2028 and 28% for 2029. Spending in 2030 is still projected at more than twice the 2026 level after the cut, so what the revision describes is slower growth, not a contraction. A separate estimate covering total accelerator silicon area shipping from 2026 to 2030 was reportedly reduced by 17%, with the Market Leader responsible for about two-thirds of that reduction. The model is sold only to institutional subscribers, so the numbers reached the market secondhand, which helps explain an abrupt reaction in place of a gradual one.

The Reported Reasons for the Change

The summaries named three reasons. Slower Revenue growth at frontier AI labs ranks first and matters most; net new annualised Recurring Revenue there reportedly decelerated in the third quarter of 2026. Tighter financing for data centers is second, especially for projects lacking long-term offtake agreements from large cloud providers. Third, and described as a secondary headwind, are delays to US Data Center capacity of about 5.8 gigawatts, tied to regulatory and political changes following the midterm elections. Taken together, these describe a market in which Demand keeps rising but funding and power, more than chip supply, increasingly set the pace. Other industry trackers put 2028 Capital spending far higher, so the forecast stands in contrast to them, and the spread of views on AI infrastructure is plainly wide. What matters to chip investors is the shape of the curve, not its absolute level.

Consequences for Semiconductor Shares

Because the most valuable chip companies are priced for years of strong growth, the market reacted sharply. Micron fell 7.6%, AMD 5.6%, Nvidia 4.1% and Broadcom 3.9% on Monday, the iShares Semiconductor ETF shed 5.37%, and a three times leveraged chip fund sank by more than 16%. Asian chipmakers had already been hit hard, with Taiwan's main index down 6.03%. Tuesday, when Nvidia reports, brings the real test. Whether the lower forecast is supported or challenged will depend on its outlook for the January quarter and any remarks about 2027 orders. Concerns would probably ease on strong commentary from the largest buyer group, whereas hesitation would lend the revised numbers more weight. Concentration is a key Factor for SMH holders: the fund's largest positions are the companies most exposed to accelerator demand, so its performance will follow that debate closely over the coming weeks.

Where Things Stand

Monday's chip sell-off was set off by a reported forecast cut and not by a company warning. Unconfirmed figures remain the basis, and Nvidia's November 17 results Supply the first hard data point to test whether AI accelerator spending is slowing as the revised model suggests.