World Semiconductor Trade Statistics (WSTS), “Global semiconductor market nears $1 trillion in 2026», December 2, 2025.
A Deloitte study (Spring 2026) on the AI chip market initially estimated that AI chips in 2026 would be worth approximately US$300 billion. Given the World Semiconductor Trade Statistics’ December 2025 upward revision of the US$175 billion global chip market (all driven by demand for AI, with weakness in non-AI markets), Deloitte now estimates that the AI chip market in 2026 will reach approximately US$500 billion.
Deloitte analysis based on quarterly earnings discussions, presentations and insights shared by leading semiconductor companies, cross-validated with secondary research data from third-party industry sources such as WSTS and International Data Corporation (IDC).
Deloitte analysis of public market capitalizations in 2025, 2024 and 2023. The 2025 market capitalization was last updated based on the market close of December 15, 2025.
See note 2.
Manufacturing AND“AMD predicts AI chip market to exceed $500 billion by 2028“, June 13, 2025; Max A. Cherney, “AMD expects profits to triple by 2030 and data center chip market to reach $1 trillion», Reuters, November 12, 2025.
Deloitte analysis based on: Semiconductor Industry Association (SIA), “State of the US Semiconductor Industry 2025“, July 2025; based on our discussions with SIA, unit sales for 2025 (as of December 16, 2025) were 10% higher than 2024, and SIA estimates that total chip sales for all of 2025 will be 1.05 trillion units. Using WSTS estimates of 772 billion dollars of total semiconductor industry revenue for 2025, the average selling price per unit is $0.74.
A major AI chip maker has secured around 800,000 wafers for its core chip in 2026 and is producing around 20 chips per wafer, suggesting around 16 million chips in total. With a market share of around 80%, we estimate that the total annual number of AI chips produced could be around 20 million.
SEMI, “SEMI Announces Global Silicon Wafer Shipments to Rebound 5.4% in 2025, With New Record Expected by 2028», press release, October 28, 2025.
The times of the straits, “Smartphone and PC sales expected to fall as prices rise due to rising memory chip costs“, January 22, 2026; Francisco Jeronimo, “Global Memory Shortage Crisis: Market Analysis and Potential Impact on Smartphone and PC Markets in 2026“, IDC, December 18, 2025; Dan Robinson, “Budget smartphones will be hit hardest by rising memory prices“, The registerJanuary 15, 2026.
TrendForce, “The rise in memory prices is weighing on consumer markets; Smartphone and laptop outlook for 2026 revised down, says TrendForce», press release, November 17, 2025.
WSTS, “Global Semiconductor Market Shows Continued Growth in Q2 2025», press release, August 4, 2025.
Luke James, “AI data centers are gobbling up the world’s supply of memory and storage, setting the stage for a pricing apocalypse that could last a decade.», Tom’s Hardware, October 3, 2025.
PCPartPicker, “Memory Price Trends», accessed on December 18, 2025.
Counterpoint, “Memory prices climb 50% in the fourth quarter and the rise will continue in 2026», January 7, 2026.
See note 2.
Goldman Sachs,How AI is transforming data centers and increasing energy demand», August 29, 2025.
Luke James, “Jet engine shortages threaten AI data center expansion as wait times extend to 2030 – rush to develop AI continues», Tom’s Hardware, October 27, 2025.
Nestor Maslej et al., “The AI Index 2025 annual report», Human-centered AI from Stanford University, April 2025.
Based on Deloitte’s analysis of articles from several third-party sources, including Semiconductor Engineering, The Wall Street Journal, Forbesand Seeking Alpha, published during the second half of 2025 and the first quarter of 2026.
Josh You and David Owen, “How much power will advanced AI training require in 2030?» Epoch AI, August 11, 2025.
Duncan Stewart, Karthik Ramachandran, Prashant Raman and Ariane Bucaille, “Rising Trends: New and Upcoming Technologies Worth Putting on Your Radar“, Deloitte InsightsNovember 19, 2024.
Deloitte Consulting LLP performed an analysis of the AI data center rack infrastructure, including a rough bill of materials of the various semiconductor-based components and the underlying dynamics of their supply chain. This analysis is expected to be published in the first quarter of 2026.
In co-packaged optics, optical transceivers are placed directly next to (or within) the same package as the application-specific processor or integrated circuit using ultra-short die-to-die interconnects, strengthening integration and, therefore, shortening the signal path between the processor and the optical interface. This helps reduce signal and electricity losses and reduces power consumption. To read further, see: Sharada Yeluri, “Co-packaged optics – a deep dive»,APNIC, May 7, 2025.
"Oscoo,"HBF: a new high-bandwidth flash star that breaks the “memory wall” for AI», November 3, 2025.
Link-PP, “AI Data Centers Drive Optical Transceiver Market Growth», December 10, 2025.
Deloitte analysis based on data from “Gartner, Forecast Analysis: AI Network Fabric, Worldwide, 2H25“, October 31, 2025. Gartner is a registered trademark of Gartner, Inc. and/or its affiliates and is used herein with permission. All rights reserved.
Yeluri, “Co-packaged optics – a deep dive.”
Based on Deloitte’s analysis of data center AI chip offerings and solutions from silicon merchant players, fabless design firms and hyperscalers.
Based on our conversations with semiconductor industry experts, we see that back-end chip manufacturing efforts are limited in Europe, likely due to the growing number of fabless companies in the region. And that could affect the region’s ability to build its end-to-end chip manufacturing capabilities, both on the front and back end. To learn more about talent challenges in the semiconductor industry, visit: The Journal of the Chronicle“The Looming Silicon Ceiling: Semiconductor Talent Shortage Threatens Global AI Ambitions», December 12, 2025.
Jeroen Kusters, Deb Bhattacharjee, Jordan Bish, Jan Thomas Nicholas, Duncan Stewart and Karthik Ramachandran, Global Semiconductor Industry Outlook 2025, Deloitte InsightsFebruary 4, 2025.
Based on Deloitte’s analysis of several strategic corporate investments made by semiconductor and technology companies, including hyperscalers, AI model companies, semiconductor fabless companies and integrated device manufacturers, as well as private equity firms in 2025.
Duncan Stewart, Jeroen Kusters, Deb Bhattacharjee, Arpan Tiwari, Girija Krishnamurthy and Karthik Ramachandran, “TMT 2026 predictions: why the next phase of AI will likely require more computing power, not less“, Deloitte InsightsNovember 18, 2025.
David Jarvis, Duncan Stewart, Nick Seeber, Gillian Crossan, Tim Bottke and Girija Krishnamurthy, «TMT 2026 Predictions: A New Era of Autonomy: Navigating Technological Sovereignty“, Deloitte InsightsNovember 18, 2025; Additionally, Deloitte analyzed several state-level strategic investments in the United States, the Middle East (especially in the Gulf region), Japan and Europe, and found that various governments were funding their respective countries or regions’ AI and data center plans and supporting local manufacturing, assembly and testing, advanced packaging, and research and development.
Karthik Ramachandran, Duncan Stewart, Jeroen Kusters, Deb Bhattacharjee, Girija Krishnamurthy and Jan Nicholas, “TMT 2026 Predictions: New technologies and familiar challenges could weaken semiconductor supply chains“, Deloitte InsightsNovember 18, 2025.
Ibid.
Reuters, “Exclusive: Nvidia plans to increase production of H200 chips due to strong Chinese demand, sources say», December 15, 2025.
TrendForce, “Memory Industry to Maintain Cautious Investment in 2026, With Limited Impact on Bit Supply Growth, TrendForce Says», press release, November 13, 2025.
By 2035, Deloitte estimates that the energy demand of AI data centers in the United States could increase more than thirtyfold, to 123 gigawatts, up from 4 gigawatts in 2024. AI data centers may require significantly more energy per square foot than traditional data centers. For more, see: Martin Stansbury, Kelly Marchese, Kate Hardin and Carolyn Amon, «Can America’s infrastructure keep pace with the AI economy?» Deloitte InsightsJune 24, 2025.