SemiAnalysis Slashes 2028 AI Accelerator Spending Forecast by 28%, Nvidia Bears Brunt
SemiAnalysis has significantly revised its forecast for global AI accelerator spending, cutting its 2028 projection from $1.054 trillion to $758 billion—a 28% reduction. The firm also slashed its 2030 revenue estimate for Nvidia’s compute products by 66%, while Broadcom’s custom chip cumulative revenue forecast saw a milder 11% cut.
The downward revision reflects growing skepticism about the sustainability of AI infrastructure spending, particularly around Nvidia’s dominant GPU business. SemiAnalysis cites slower-than-expected enterprise adoption, supply chain constraints, and increasing competition from custom ASICs as key factors. The stark difference in impact—Nvidia’s massive cut versus Broadcom’s modest reduction—highlights a shift toward customized silicon solutions from hyperscalers like Google, Amazon, and Meta.
Industry Implications
For the crypto and blockchain sector, the revised forecast has mixed implications. AI-focused crypto projects—especially those building decentralized compute networks and GPU marketplaces—may face a tighter funding environment if the AI investment boom cools. However, the pivot toward custom chips could benefit blockchain-based AI inference platforms that optimize for specific workloads. Meanwhile, Nvidia’s lowered outlook might pressure crypto mining stocks that have diversified into AI compute hosting.
Broadcom’s resilience suggests that companies with diversified custom silicon businesses are better positioned to weather a spending pullback. This could accelerate the trend of hyperscalers designing their own AI chips, potentially reducing reliance on Nvidia’s general-purpose GPUs and altering the competitive landscape for decentralized compute networks that aggregate GPU resources.
Forward-Looking Perspective
While the near-term outlook for AI accelerator spending has dimmed, the long-term trajectory remains robust, albeit at a more moderate pace. Crypto projects in the AI space should focus on cost-efficient, specialized solutions and real utility rather than speculative infrastructure buildouts. The market may see consolidation among decentralized compute providers, with those offering verifiable, high-performance inference gaining an edge. Investors should watch for further forecast revisions and how Nvidia and Broadcom adapt their strategies to the evolving demand landscape.




