AMD

AMD's Q1 2026 results comfortably exceeded expectations, with total revenue up 38% year over year driven almost entirely by the data center business. Data center revenue hit $5.8 billion, up 57% from the same quarter a year prior, ahead of analyst estimates and marking the segment's strongest quarterly growth rate in recent history.

Non-GAAP earnings per share came in at $1.37, beating consensus by roughly eight percent — a meaningful cushion that management attributed to a more favorable product mix as Instinct GPU shipments scaled. Non-GAAP gross margin expanded year over year.

Q1 free cash flow more than tripled year over year, reaching a new quarterly record. Management guided Q2 revenue to approximately $11.2 billion — well above the street — prompting a sharp aftermarket rally.

In January 2026, AMD unveiled the full Instinct MI400 series at CES, built on CDNA 5 architecture and TSMC's 2nm node, with two main variants: the MI455X for AI training and inference, and the MI430X targeting HPC and sovereign AI. The Helios rack-scale platform, combining 72 MI455X GPUs with EPYC Venice CPUs, was shown delivering roughly 2.9 exaFLOPS per rack.

AMD and OpenAI signed a multibillion-dollar, multi-year supply agreement in late 2025 covering six gigawatts of AMD GPU capacity, with first deployments targeted for the second half of 2026. A parallel expansion with Meta was announced for custom MI450-based hardware supporting Meta's first gigawatt-scale AI build.

On May 31, 2026, the US Bureau of Industry and Security extended licensing requirements to cover AMD's MI350X when sold to Chinese-controlled entities outside China, closing an indirect procurement route through third-country subsidiaries. CEO Lisa Su had traveled to Beijing earlier in May for bilateral talks with Chinese officials — an unusually direct diplomatic gesture for a chipmaker.

The bull case rests on hyperscaler customer concentration turning into a flywheel: OpenAI, Meta, Microsoft, Oracle, and Google all deepening AMD commitments simultaneously creates a credible multi-year revenue ramp that analysts argue is not yet fully in consensus estimates.

The skeptic's concern is twofold: China, which still represents roughly 20% of revenue, faces tightening export controls that could compress a meaningful portion of the top line; and Nvidia's software ecosystem (CUDA) remains deeply entrenched in the largest AI training workflows, making share gains genuinely difficult above a certain volume threshold.

AMD's PEG ratio well below 1 has become a key bull argument — suggesting the stock is cheap relative to its growth trajectory, even after a massive year-over-year run. Whether the implied forward growth materializes on schedule is the swing factor that will define the next leg of sentiment.