If you've been following the AI chip race, you know it's mostly been a one-name show. But Advanced Micro Devices, Inc. (AMD) just delivered a quarter that makes it clear: this is now a two-horse race.
Counterpoint Research analyst Neil Shah said Thursday that AMD's first-quarter 2026 results were blockbuster — revenue jumped 38% year over year to $10.3 billion, powered by strong demand for both CPUs and accelerators. The real headline, though, is the data center business: revenue there climbed 57% to $5.8 billion, now making up 56% of AMD's total revenue.
Shah argues that AMD has cemented its position as the only credible full-stack merchant alternative to NVIDIA Corp. (NVDA). That means AMD offers CPUs, GPUs, FPGAs, networking, packaging, its ROCm software, and even rack-scale solutions. It's not just a chip company anymore — it's an infrastructure company.
Of course, NVIDIA still dominates. Shah notes that NVIDIA's data center revenue is still roughly seven times AMD's. But AMD is narrowing the gap, and the rise of Agentic AI is giving it a tailwind.
Agentic AI Brings CPUs Back Into the Spotlight
One of the more interesting shifts Shah highlights is how Agentic AI — AI systems that can act autonomously — is reviving demand for CPUs. These workloads require more balanced compute architectures, not just a pile of GPUs. That's good news for AMD's EPYC line.
Shah says AMD is seeing strong traction for its 5th-Gen EPYC processors across cloud and edge deployments. The company expects CPU revenue to rise 70% year over year next quarter. And the upcoming 6th-Gen Zen 6 EPYC chips, built on Taiwan Semiconductor Manufacturing Co. Ltd.'s (TSM) 2nm process, could drive further growth.
NVIDIA Still Leads, But AMD Is Closing
Shah describes NVIDIA and AMD as a "two-horse race" in full-stack AI infrastructure. At NVIDIA's GTC conference, AI inferencing took center stage, with growing attention on token economics and efficient compute architectures. NVIDIA continues to benefit from the broader AI infrastructure boom as hyperscalers and enterprises expand investments.
But the real battle is next year. Shah says the competition between AMD's MI400 series and NVIDIA's Rubin Ultra platform will be a key battleground. Both companies are pushing the envelope on performance and efficiency.
Intel Stumbles, ARM Emerges
AMD's gains aren't just about its own execution. Shah points out that Intel Corp. (INTC) has had supply issues and has been unable to meet CPU demand. That's helped AMD grab server CPU market share, and Shah says AMD could potentially surpass Intel in x86 server CPUs.
ARM-based server CPUs are also emerging as a strong alternative, though x86 still dominates. But the shift is real, and it's adding more competition to the market.
The Ecosystem Matters
Beyond the chipmakers themselves, the AI supply chain is complex. Shah highlights Taiwan Semiconductor as a critical enabler, with competition for its N2, N3, and CoWoS capacity set to intensify this year. Samsung Electronics Co., Ltd. (SSNLF) is also a key player, with its HBM4 collaboration with AMD being an important memory partnership for future AI accelerators.
On the customer side, Shah names Meta Platforms, Inc. (META), OpenAI, Anthropic, Amazon.com, Inc. (AMZN) (Amazon Web Services), Alphabet Inc. (GOOGL) (Google Cloud), Microsoft Corp. (MSFT) (Azure), and Tencent Holding Ltd. (TCEHY) as major AI infrastructure customers driving semiconductor demand. Downstream, Wistron Corp and Foxconn Industrial Internet are important manufacturing partners supporting gigawatt-scale AI deployments.
Price Action
As of Thursday premarket trading, NVIDIA shares were up 0.47% at $208.80, while AMD shares were down 0.88% at $417.70.