HaiAI123

Curated Global AI Tools Directory

Industry update
Flat

SemiAnalysis Cuts AI Compute Forecast, Trims 2030 Capex by $608 Billion

11/16/2026 — 11/16, 21:21·3 sources (all)·3 reports (all)

Key takeaways

  • SemiAnalysis cut its AI compute forecast in a November 13 model update: 2028 global AI accelerator capex is now 28% lower, while 2029 and 20…
  • Nvidia accounts for 66% of the reduction in silicon area shipments.

Background & analysis

Semiconductor research firm SemiAnalysis released a model update on November 13 that scaled down its AI compute forecasts, covering AI accelerator shipments, capital spending, HBM demand and power demand. In the revision, the 2028 forecast for global AI accelerator capital expenditure was cut by 28%, with 2029 and 2030 lowered by 41% and 40% respectively.

Under the new model, 2030 global AI accelerator capital expenditure falls from the prior projection of 1.531 trillion dollars to 923 billion dollars, a reduction of 608 billion dollars. Cumulative forecasts for accelerator logic silicon shipment area from 2026 through 2030 are down 17%, and NVIDIA accounts for 66% of that reduction. One report adds that SemiAnalysis also sharply lowered its shipment estimates for NVIDIA over the coming years and reduced its average selling price assumptions for the company's next-generation products, suggesting NVIDIA would need to accept thinner margins to stay competitive. The 2026 capital expenditure forecast was left largely unchanged, still standing at 392 billion dollars. Both reports simply relay the revised figures, and the event currently remains at the stage of a downward revision to the forecasts.

AI-generated from 2 reports · updated 16 hours ago

Latest turnSemiconductor research firm SemiAnalysis sharply cut its forecasts for AI accelerator shipments, capex, HBM and power demand. Its 2028 global AI accelerator capex estimate was lowered 28%, with 2029 and 2030 down 41% and 40%, pulling the 2030 figure to $923 billion from $1.531 trillion. Cumulative accelerator logic wafer area forecasts for 2026-2030 were cut 17%.

updream
24-hour heatHeat index 184 · peak 298 · 16h ago
24 hours agonow

Reports on this story headlines open the original

Yesterday
  1. Semiconductor research firm SemiAnalysis sharply cut its forecasts for AI accelerator shipments, capex, HBM and power demand. Its 2028 global AI accelerator capex estimate was lowered 28%, with 2029 and 2030 down 41% and 40%, pulling the 2030 figure to $923 billion from $1.531 trillion. Cumulative accelerator logic wafer area forecasts for 2026-2030 were cut 17%.

    36氪AI score 64
  2. SemiAnalysis has sharply cut its forecasts for AI accelerator shipments, capex, HBM demand and power demand in its latest model update. It now expects global AI accelerator capex in 2028 to be 28% lower than previously projected, with 2029 and 2030 cut by 41% and 40%. The 2030 capex forecast drops from $1.531 trillion to $923 billion.

    东方财富AI score 64
  3. Semiconductor research firm SemiAnalysis cut its AI accelerator forecasts in a November 13 model update, lowering projected 2028 capex by 28% and the 2029 and 2030 figures by 41% and 40% respectively. It now expects 2030 capex of $923 billion, down from $1.531 trillion, with Nvidia accounting for 66% of the reduction in silicon area shipped.

    华尔街见闻AI score 68

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →