NVIDIA reported $75.2 billion in data center revenue for the quarter ended April 26, 2026, up 92% year over year. AMD’s entire data center segment did $5.8 billion in its most recent quarter. Intel’s did $5.1 billion, most of it from CPUs, not GPUs. This post breaks down AI GPU market share across NVIDIA, AMD, and Intel with the latest 2026 revenue figures, share estimates, and margin data.
AI GPU Market Share: Key Facts
- NVIDIA holds an estimated 75% to 81% of AI accelerator revenue in 2026, based on Silicon Analysts and IDC estimates.
- AMD’s Instinct GPU line generated an estimated $7 to $8 billion in 2025, roughly 5% to 7% of the market.
- Intel’s AI GPU share sits near 1%, with its Crescent Island GPU not sampling until the second half of 2026.
- The 2026 data center accelerator market exceeds $200 billion, per Silicon Analysts.
- Custom ASIC shipments are growing 44.6% in 2026 versus 16.1% for merchant GPUs, according to TrendForce.
The short version: NVIDIA still controls roughly three quarters of AI GPU revenue in 2026, but its share has fallen from a peak near 87% in 2024. AMD is now a credible second source with $5.8 billion in quarterly data center revenue and 6-gigawatt deals with both Meta and OpenAI. Intel remains a rounding error in AI GPUs while it prepares a 2026 relaunch.
Who Leads the AI GPU Market in 2026?
NVIDIA leads by a wide margin. Analyst estimates for its 2026 AI accelerator revenue share range from 70% to 86% depending on whether the count includes hyperscaler custom silicon.
AMD holds roughly 5% to 7%. Intel holds about 1%, and custom chips from Google, Amazon, Microsoft, and Meta account for most of the remainder.
| Vendor | 2026 Revenue Share (Est.) | Primary AI Products |
|---|---|---|
| NVIDIA | 75-81% | Blackwell B200/B300, GB300 NVL72 |
| AMD | 5-7% | Instinct MI350/MI355X, MI450 |
| Intel | ~1% | Gaudi 3, Crescent Island (H2 2026) |
| Custom ASICs and others | 12-18% | Google TPU, AWS Trainium, Maia, MTIA |
Source: Silicon Analysts, IDC, TrendForce estimates, Q1-Q2 2026
No audited market share table exists for AI accelerators. Revenue share flatters NVIDIA because its chips carry premium pricing, while a TPU or Trainium chip is an internal cost item for Google or Amazon.
NVIDIA AI GPU Market Share and Revenue in 2026
NVIDIA’s fiscal 2026 revenue reached $215.9 billion, up 65% from the prior year, with data center sales above 90% of the total. Data center revenue for the year came to $193.7 billion, based on Silicon Analysts tallies of quarterly filings.
The most recent quarter, ended April 26, 2026, set new records: $81.6 billion in total revenue, $75.2 billion from data center, and $14.8 billion from data center networking alone, up 199% year over year.
Share is a different story. NVIDIA’s slice of the AI accelerator market peaked near 87% in 2024 and is projected at roughly 75% for 2026, per Silicon Analysts. IDC pegs it at 81%.
| Fiscal Year | NVIDIA Data Center Revenue | Est. Accelerator Share |
|---|---|---|
| FY2024 | $47.5B | ~87% |
| FY2025 | $115.2B | ~80% |
| FY2026 | $193.7B | ~75-80% |
Source: NVIDIA quarterly filings, Silicon Analysts
CUDA remains the moat. More than 5 million developers build on it, which keeps training workloads locked to NVIDIA hardware even where AMD matches raw FP8 throughput. Nearly all of that hardware runs on Linux, which continues to dominate server market share across industry sectors.
How Much AI GPU Market Share Does AMD Hold?
AMD’s data center segment recorded $5.8 billion in Q1 2026, up 57% year over year, driven by EPYC CPUs and Instinct MI350 series GPUs. Full-year 2025 data center revenue was $16.6 billion.
The Instinct GPU line alone generated an estimated $7 to $8 billion in 2025, about 5% to 7% of AI accelerator revenue. Q4 2025 included roughly $390 million of MI308 GPU sales to China.
AMD’s 2026 pipeline is what changed the conversation. Two 6-gigawatt GPU commitments now anchor its roadmap.
| Deal | Scale | Announced | First Shipments |
|---|---|---|---|
| OpenAI | 6 GW of Instinct GPUs | October 2025 | H2 2026 |
| Meta | Up to 6 GW, first 1 GW custom MI450 | February 2026 | H2 2026 |
Source: AMD Q1 2026 earnings, Datacenter Dynamics
AMD expects initial volume of its Helios rack-scale systems in Q3 2026, with a larger ramp in Q4 and into early 2027. The company guided Q2 2026 revenue to $11.2 billion, plus or minus $300 million. Enterprises weighing these deployments face the same AI governance questions that hardware alone does not solve.
Intel AI GPU Market Share: Where Does It Stand?
Intel’s Data Center and AI segment posted $5.1 billion in Q1 2026, up 22% year over year, but the growth came from Xeon CPUs handling agentic AI workloads, not GPUs. Its Gaudi 3 accelerator missed the company’s own $500 million revenue target for 2024 and never recovered.
The next attempt is Crescent Island, a data center GPU built on the Xe3P architecture with 160GB of LPDDR5X memory, designed for air-cooled inference racks. Customer sampling starts in the second half of 2026, which means meaningful revenue lands in 2027 at the earliest.
Intel stock is still up more than 190% year to date in 2026 on foundry momentum and CPU inference demand. NVIDIA itself invested $5 billion in Intel in September 2025.
NVIDIA vs AMD vs Intel: Latest Quarterly Numbers
The revenue gap remains stark. NVIDIA’s data center business is roughly 13 times AMD’s and 15 times Intel’s on a quarterly basis.
| Company | Segment | Latest Quarter | Revenue | YoY Growth |
|---|---|---|---|---|
| NVIDIA | Data Center | Q1 FY2027 (Apr 2026) | $75.2B | +92% |
| AMD | Data Center | Q1 2026 (Mar 2026) | $5.8B | +57% |
| Intel | Data Center & AI | Q1 2026 (Mar 2026) | $5.1B | +22% |
Source: NVIDIA, AMD, and Intel quarterly earnings reports, 2026
Growth rates tell a second story. AMD is compounding faster than Intel, and NVIDIA is compounding faster than both despite its size.
Chip-Level Gross Margins Compared
Margin structure explains why NVIDIA can outspend rivals on R&D and TSMC capacity. Silicon Analysts estimates chip-level gross margins of roughly 88% on the H100 and 84% on the B200, against 64% to 68% for AMD’s MI300X and MI355X and 58.4% for Intel’s Gaudi 3.
NVIDIA’s corporate GAAP gross margin ran 74.9% in its latest quarter. AMD reported 53% GAAP and 55% non-GAAP in Q1 2026.
AI GPU Market Share Under Pressure From Custom Silicon
The fastest-growing competitor to all three GPU vendors is not a GPU. TrendForce data from May 2026 shows custom ASIC shipments growing 44.6% year over year, nearly triple the 16.1% rate for merchant GPUs.
ASIC-based servers are on track for 27.8% of accelerator server shipments in 2026, the highest share since 2023. Broadcom, the dominant custom chip architect, carries a $73 billion AI backlog and targets $100 billion in annual AI revenue by 2027.
The displacement is concentrated in inference, where workloads are predictable and volumes justify custom design. Much of that inference is also moving closer to users, a shift visible in Linux edge computing deployment statistics.
Training stays with NVIDIA. Anthropic’s commitment of up to one million Google TPUs and Microsoft running GPT-5.2 workloads on Maia show ASICs proven at frontier scale, yet CUDA’s flexibility keeps diverse enterprise workloads on merchant GPUs.
Storage and data pipelines scale alongside the accelerators; object storage deployment on Linux platforms tracks that build-out. For a wider view of where all this compute is heading, see these 2026 AI trends and predictions.
What the 2026 Numbers Add Up To
The market is expanding faster than NVIDIA’s share is shrinking. A 75% slice of a $200 billion-plus market in 2026 is worth far more than an 87% slice of the 2024 market was.
AMD has secured its position as the merchant second source, with real hyperscaler commitments and a rack-scale product in Helios. Intel’s AI GPU business is effectively a 2027 story that depends on Crescent Island execution.
The variable to watch is not AMD versus NVIDIA. It is how much inference spending hyperscalers divert to their own silicon.
FAQs
Who has the biggest AI GPU market share in 2026?
NVIDIA leads with an estimated 75% to 81% of AI accelerator revenue in 2026, per Silicon Analysts and IDC. AMD holds 5% to 7% and Intel about 1%, with custom hyperscaler chips taking most of the rest.
How much data center revenue does NVIDIA make?
NVIDIA reported $75.2 billion in data center revenue for the quarter ended April 26, 2026, up 92% year over year. Full fiscal 2026 data center revenue reached $193.7 billion.
Is AMD gaining AI GPU market share?
Yes. AMD’s data center revenue grew 57% year over year to $5.8 billion in Q1 2026, and it holds 6-gigawatt GPU deals with both OpenAI and Meta, with MI450 shipments starting in the second half of 2026.
Does Intel compete in AI GPUs?
Barely, for now. Gaudi 3 missed its $500 million 2024 target, and Intel’s share is near 1%. Its Crescent Island GPU begins customer sampling in the second half of 2026, targeting inference workloads.
How big is the AI accelerator market in 2026?
Silicon Analysts pegs the 2026 data center accelerator market above $200 billion, up from roughly $55 billion in 2023. Inference is on track to represent about two thirds of all spending.
Sources
https://www.sec.gov/Archives/edgar/data/0001045810/000104581026000051/q1fy27pr.htm
