AI Chip Statistics 2026: Market Size, Vendors and Supply

NVIDIA posted Data Center revenue of $39.1 billion in the first quarter of fiscal 2026, a single-quarter figure larger than any AI chip rival’s full-year sales. The figures below draw on primary issuer filings, SIA data, and US Commerce Department disclosures.

Key Takeaways

• Global semiconductor sales reached $791.7 billion in 2025, an increase of 25.6% compared to the prior year’s $630.5 billion, per the Semiconductor Industry Association’s February 6, 2026, release.
• Logic chips, the category that contains AI accelerators, grew 39.9% in 2025 to $301.9 billion, making them the largest semiconductor product category by sales.
• NVIDIA’s quarterly Data Center revenue rose 73% year-over-year to $39.1 billion in Q1 fiscal 2026, the line item driving most of the AI silicon market.
• AMD’s Q1 2026 Data Center segment reached $5.8 billion, up 57% year-over-year, driven by EPYC processors and Instinct GPU shipments.
• Broadcom reported $8.4 billion in AI semiconductor revenue in Q1 fiscal 2026, a 106% year-over-year increase, with custom XPU silicon as the primary growth engine.
• TSMC’s high-performance computing segment, which includes AI accelerators, accounted for a record high of 61% of revenue, with quarter-over-quarter growth of approximately 20%.
• The US Commerce Department had awarded approximately $33.7 billion of CHIPS Act direct funding as of January 31, 2025, with Intel’s $7.86 billion the single largest grant.

Editor’s Choice

• The global semiconductor industry is projected to reach roughly $1 trillion in worldwide annual sales in 2026, per SIA forecasts.
• NVIDIA’s total Q1 fiscal 2026 revenue hit $44.1 billion, up 69% year-over-year.
• Hyperscalers (Microsoft, Google, Meta, Amazon) are on track to spend approximately $650 billion on AI infrastructure in 2026, up about 80% from the prior year.
• Broadcom’s named custom-silicon customer list grew from 5 to 6 hyperscalers in Q1 fiscal 2026, with OpenAI joining the named cohort.
• AMD guided Q2 2026 revenue to approximately $11.2 billion plus or minus $300 million, a 46% year-over-year increase at the mid-point.
• TSMC raised its full-year 2026 revenue growth forecast from 30% to over 30%, with capex guided toward the high end of its $52.0 to $56.0 billion range.
• US semiconductor-related startup equity funding totaled $6.2 billion in 2025, up 85% year-over-year, the highest annual tally on record.

Recent Developments

• May 5, 2026: AMD reported first-quarter revenue of $10.3 billion, gross margin of 53%, and net income of $1.4 billion, with Data Center now the company’s largest segment.
• April 16, 2026: TSMC released Q1 2026 results showing US-dollar revenue of $35.89 billion, up 40.6% year-over-year, with HPC at a record-high 61% of revenue.
• March 6, 2026: Broadcom announced Q1 fiscal 2026 AI semiconductor revenue of $8.4 billion, up 106% year-over-year, and guided Q2 to $10.7 billion (a 140% year-over-year jump).
• February 6, 2026: The Semiconductor Industry Association reported Q4 2025 chip sales of $236.6 billion, 37.1% above Q4 2024, capping a record full year.
• April 9, 2025 (still active): The US government’s H20 export licensing requirement triggered a $4.5 billion NVIDIA inventory charge and a $2.5 billion unshipped revenue gap in Q1 fiscal 2026; the regulatory regime remained in force through Q1 2026 publication.

AI Chip Market Size and Revenue

• Global semiconductor sales totaled $791.7 billion in 2025, setting a record.
• Annual sales are projected to reach roughly $1 trillion in 2026, per SIA forecasts.
• Logic chips grew 39.9% in 2025 to $301.9 billion, the largest product category.
• Memory products increased 34.8% to $223.1 billion in 2024.
• Q4 2025 alone hit $236.6 billion, up 13.6% sequentially.
• Asia Pacific (excluding Japan) grew 45.0% year-over-year in 2025, the fastest region.
• The Americas grew 30.5% in 2025, driven by AI infrastructure buildout.
• Hyperscaler 2026 AI capex is on track for roughly $650 billion, up about 80% year-over-year.
• Logic chip growth outpaced the broader industry in 2025, signalling that AI accelerators captured most of the incremental spending.

Source: Semiconductor Industry Association

Subscribe To Our Newsletter!

Be the first to get exclusive offers and the latest news.

Email address

Subscribe Now

NVIDIA’s Data Center Revenue and Market Position

• NVIDIA’s Q1 fiscal 2026 total revenue was $44.1 billion, up 12% sequentially and 69% year-over-year.
• Data Center revenue reached $39.1 billion, up 73% year-over-year.
• Gaming revenue hit a record $3.8 billion, up 42% year-over-year.
• GAAP gross margin was 60.5%, with non-GAAP at 61.0%.
• Excluding the H20 charge, non-GAAP gross margin would have been 71.3%.
• NVIDIA captured approximately 86% market share in AI accelerator sales in 2025, roughly unchanged from 2024.
• Q2 fiscal 2026 guidance was set at approximately $45.00 billion, absorbing an estimated $8.00 billion H20 revenue loss.
• Per NVIDIA’s Q1 fiscal 2026 release, Data Center revenue of $39.1 billion was up 10% sequentially and 73% year-over-year, with further context on the company’s workforce in how many people work at NVIDIA.

Source: NVIDIA Investor Relations

By the numbers: NVIDIA’s $39.1 billion Q1 fiscal 2026 Data Center line outpaced AMD’s $5.8 billion Q1 2026 Data Center revenue by roughly 6.7 times, per each company’s earnings release. The gap reflects NVIDIA’s incumbent CUDA stack plus priority TSMC CoWoS allocation that AMD’s MI350 ramp has not yet closed.

AMD’s Data Center Segment and MI-Series Ramp

• AMD’s Q1 2026 Data Center segment revenue was $5.8 billion, up 57% year-over-year.
• Total Q1 2026 revenue was $10.3 billion, with non-GAAP diluted EPS of $1.37.
• Client and Gaming segment revenue reached $3.6 billion, up 23% year-over-year.
• Non-GAAP gross margin was 55% on a non-GAAP basis, according to the company’s earnings release.
• AMD guided Q2 2026 revenue to approximately $11.2 billion, plus or minus $300 million.
• The Q2 mid-point represents approximately 46% year-over-year revenue growth.
• Growth was driven by demand for AMD EPYC processors and the continued ramp of AMD Instinct GPU shipments, per CEO Lisa Su.
• Dr. Lisa Su, chair and CEO of AMD, said the company delivered an outstanding first quarter, driven by accelerating demand for AI infrastructure, with Data Center now the primary driver of our revenue and earnings growth.

Why are AMD’s Instinct accelerators gaining traction with hyperscalers?

AMD’s Q1 2026 Data Center revenue of $5.8 billion, up 57% year-over-year, reflects strong demand for EPYC processors and the continued ramp of Instinct GPU shipments, per the company’s earnings release. Hyperscalers needed a credible second source amid NVIDIA Blackwell supply constraints, which accelerated the Instinct ramp.

Broadcom Custom Silicon and Hyperscaler XPU Customers

• Broadcom Q1 fiscal 2026 AI semiconductor revenue was $8.4 billion, up 106% year-over-year.
• Q2 fiscal 2026 AI revenue guidance is $10.7 billion, a 140% year-over-year increase.
• The named XPU custom-silicon customer count expanded from 5 to 6 in Q1 fiscal 2026, with OpenAI joining Google, Anthropic, Meta, and two unnamed hyperscalers.
• Broadcom reported a $73 billion backlog of committed customer orders for AI silicon.
• Broadcom CEO Hock Tan stated the company has a line of sight to achieve AI revenue from chips in excess of $100 billion in 2027, supported by the $73 billion backlog.
• The hyperscaler appetite for non-NVIDIA custom silicon, expressed through Broadcom’s XPU customer count, is the clearest signal that the largest AI buyers are working to escape single-vendor concentration risk on their compute budgets.
• AI semiconductor gross margins on Broadcom’s custom-XPU business reflect custom-silicon pricing power, though the company does not break out segment-level gross margin disclosures separately from group results.

Source: Broadcom Investor Relations

TSMC Foundry Capacity and Advanced-Node Mix

• TSMC Q1 2026 US-dollar revenue was $35.89 billion, up 40.6% year-over-year.
• Local-currency revenue reached TWD 1.13 trillion, up 35.1% year-over-year.
• HPC accounted for a record high of 61% of revenue, with quarter-over-quarter growth of 20%.
• Advanced nodes (7nm and below) made up 74% of wafer revenue in the quarter.
• 3nm contributed 25% of wafer revenue.
• 5nm contributed 36% of wafer revenue.
• Gross margin rose to 66.2% from 62.3% in the prior quarter.
• AI accelerator revenue is tracking a mid- to high-50% CAGR through 2029, per TSMC management.
• 2026 capex is guided toward the high end of its $52.0 to $56.0 billion range.

Hyperscaler AI Capital Expenditure

• Microsoft, Google, Meta, and Amazon placed multi-billion-dollar forward orders for Blackwell GPUs in 2025, consuming most of NVIDIA’s available allocation through 2026 and into 2027.
• Combined hyperscaler AI capex is on track for approximately $650 billion in 2026, up about 80% year-over-year, mirroring the surge in enterprise generative AI adoption.
• NVIDIA has projected approximately $1 trillion in confirmed AI chip demand through 2027 based on hyperscaler purchase orders.
• Blackwell systems were completely sold out through the middle of 2026 as of April 2026.
• Data-center GPU lead times stretched to 36 to 52 weeks across the cohort.
• Workstation GPU lead times ran 12 to 20 weeks.
• Order concentration across four hyperscaler buyers gives NVIDIA unusual revenue visibility for a hardware vendor.

Source: Center for a New American Security, Bloomberg memory-chip-shortage analysis

AI Chip Market Share by Vendor

• NVIDIA’s AI accelerator market share was approximately 86% in 2025, unchanged from 2024.
• AMD’s Q1 2026 Data Center segment revenue of $5.8 billion represents the clearest revenue-side challenger.
• Broadcom’s $8.4 billion Q1 FY26 AI semiconductor revenue reflects custom silicon outside the merchant GPU market.
• AMD reportedly secured a $60 billion deal with Meta for MI400 series deployment, a major hyperscaler breakthrough.
• NVIDIA’s Data Center revenue accounts for roughly three-quarters of the combined Q1 2026 AI-segment disclosures across the publicly reporting trio.
• NVIDIA’s CUDA stack and priority TSMC CoWoS allocation are the durable moats, weighed against AMD’s inference price-performance and Broadcom’s custom-silicon margins.
• The hyperscaler push to custom ASICs (Google TPU, Amazon Trainium, Meta MTIA) sits outside the merchant-GPU share data, captured indirectly via Broadcom’s named XPU customer expansion to 6 hyperscalers.

Why it matters: The 86% AI accelerator share NVIDIA holds is unchanged year-over-year despite AMD’s MI350 ramp and Broadcom’s XPU customer expansion, per Silicon Analysts data corroborated by Q1 2026 issuer revenue. Share stability in a market growing 50%+ annually means the absolute revenue gap is widening even as competitors gain percentage points elsewhere.

HBM Memory as the Binding Supply Constraint

• HBM demand is growing at roughly 80 to 100% per year, driven by AI accelerator deployments.
• HBM supply is growing at 50 to 60% per year as manufacturers add capacity.
• The supply gap is not projected to fully close until 2028 to 2029 at current investment rates.
• HBM now takes 23% of DRAM wafers, up from approximately 5% in 2023.
• TSMC’s CEO stated in November 2025 that advanced-process demand was running roughly three times ahead of what the company could produce.
• Microsoft, Google, Meta, and Amazon consumed most of NVIDIA’s allocation capacity through the end of 2026 and into 2027 via multi-billion-dollar Blackwell forward orders.
• HBM packaging and DRAM wafer dedication are the bottlenecks for AI chips rather than logic silicon. Every percentage point of DRAM capacity reallocated from consumer memory to HBM lifts pricing across the consumer-electronics supply chain.

Source: Bloomberg memory chip analysis, CNAS American AI report

Why are AI chips in short supply in 2026?

The bottleneck is HBM memory and advanced packaging, not the chips themselves. HBM consumes 23% of DRAM wafer capacity in 2026, up from approximately 5% in 2023, with demand growing 80 to 100% annually against supply growth of 50 to 60%, per Bloomberg’s analysis. TSMC’s CEO described demand as running roughly three times ahead of capacity in November 2025, leaving Blackwell systems sold out through the middle of the year.

Edge AI Chip Adoption in Smartphones and Consumer Devices

• Apple’s M4 chip includes a 16-core Neural Engine capable of 38 TOPS of AI performance.
• The iPhone 16 family’s A18 Pro chip delivers 35 TOPS of on-device AI while consuming 20% less power versus the prior generation.
• Qualcomm’s Snapdragon 8 Gen 3 delivers 45 TOPS of AI performance via integrated Neural Processing Units.
• Edge inference is where Google’s TPU v6 Trillium hits over 3x higher relative inference throughput for Stable Diffusion XL versus TPU v5e.
• Trillium TPUs are over 67% more energy-efficient than TPU v5e.
• Apple and Qualcomm dominate edge-AI silicon design, a segment where NVIDIA’s data-center scale has limited carry-over.

Source: Apple, Qualcomm, Google product disclosures

Automotive AI Chips and ADAS Compute

• The autonomous-driving chip market was estimated at $24.22 billion in 2024.
• The 2025 market size is projected at $29.73 billion, growing to $191.07 billion by 2034, a 23% CAGR.
• The narrower automotive AI processor market reached $5.6 billion in 2024 and is projected to reach $6.3 billion in 2025.
• NVIDIA’s Orin SoC delivers 254 TOPS for L2/L3 ADAS systems.
• NVIDIA’s forthcoming Thor platform targets 2,000 TOPS for Level 4 applications.
• Mobileye’s EyeQ6 delivers 34 TOPS, with the EyeQ Ultra targeting 176 TOPS.
• Mobileye’s installed base exceeds 100 million vehicles, a data advantage for AI training.
• The top 5 SoC players (Mobileye, NVIDIA, Qualcomm, Horizon Robotics, Huawei) are projected at 78% market share by 2035, up from 69% in 2025.
• Robotics platforms increasingly share automotive AI accelerator architectures; see related Robotics for the cross-segment overlap.

AI Chip Startup Funding

• US semiconductor-related startup equity funding totaled $6.2 billion in 2025, up 85% year-over-year.
• Cerebras Systems raised a $1.1 billion in Series G funding in September 2025.
• Cerebras followed with a $1 billion Series H in February 2026 at a $23 billion valuation.
• Cerebras has filed to go public in 2026.
• Tenstorrent closed a $693 million Series D led by Samsung Securities.
• Groq and SambaNova each surpassed $1.5 billion in total equity funding by early 2026.
• The 2025 funding tally tells a barbell story. Established names (Cerebras, Tenstorrent, Groq, SambaNova) absorbed most of the capital, while early-stage AI-chip-design startups remain capital-light relative to data-center AI demand.

Source: Crunchbase News semiconductor funding roundup

CHIPS Act Funding and Manufacturing Reshoring

• The CHIPS and Science Act allocates $52 billion in subsidies and tax credits for chip manufacturing companies.
• The Commerce Department’s CHIPS Program Office had awarded approximately $33.7 billion of direct funding as of January 31, 2025.
• Intel reached an agreement on up to $7.86 billion in direct funding for commercial semiconductor manufacturing projects across Arizona, New Mexico, Ohio, and Oregon.
• Micron Technologies announced a federal CHIPS and Science Act grant of $6.1 billion for fabs in Clay, New York, and Boise, Idaho.
• TSMC’s Phoenix operations received $6.6 billion in direct CHIPS Act funding plus $5 billion in low-cost loans.
• Companies in the semiconductor ecosystem have announced over 140 projects across 30 states, totaling more than $645.3 billion in announced investment.
• The CHIPS Act’s company-specific awards skew heavily toward leading-edge logic (Intel, TSMC) and memory (Micron). HBM’s packaging capacity, the actual bottleneck, receives less direct subsidy than its strategic role would suggest.

Source: US Department of Commerce, Intel Newsroom

Export Controls and Geopolitical Exposure

• On April 9, 2025, the US government informed NVIDIA that a license is required for exports of its H20 products into the Chinese market.
• NVIDIA’s resulting first-quarter H20 charge totaled $4.5 billion for excess inventory and purchase obligations.
• H20 product sales reached $4.6 billion in Q1 fiscal 2026 prior to the new export licensing requirements.
• NVIDIA was unable to ship an additional $2.5 billion of H20 revenue in Q1 fiscal 2026.
• The Q2 fiscal 2026 revenue guidance absorbs an estimated $8.00 billion H20 revenue loss from export controls, with non-GAAP gross margin projected to recover to 72.0%.
• US AI chip export controls cover NVIDIA Blackwell GB200 and B200 GPU configurations when destined for restricted markets.

The takeaway: A single April 9, 2025 export-licensing requirement removed roughly $15.1 billion of NVIDIA H20 product revenue across Q1 and Q2 fiscal 2026 through inventory charges, unshipped revenue, and guidance impact, per the company’s earnings disclosures. The single-product, single-regulator exposure is the largest disclosed AI chip regulatory event of the modern cycle.

Conclusion

NVIDIA’s $39.1 billion Q1 fiscal 2026 Data Center revenue, AMD’s $5.8 billion Data Center segment, and Broadcom’s $8.4 billion AI semiconductor revenue together capture the modern AI chip cycle in a single quarter. The combined $53.3 billion of disclosed Q1 2026 AI silicon revenue across the three issuers sits against a record $791.7 billion in total 2025 semiconductor sales and an SIA-forecast $1 trillion for this year.

Lead times for data-center GPUs range from 36 to 52 weeks, with Blackwell sold out through mid-year. The constraints that will shape this year are HBM wafer allocation, the $4.5 billion H20 charge that signaled how fast export controls can reset a quarter, and the $33.7 billion CHIPS Act allocation routing leading-edge capacity to US soil. The vendors that capture the next two years of AI chip share will secure HBM, advanced packaging, and politically resilient supply chains alongside the silicon design.