Company Insights

AMD customer relationships

Explore interactive graphBrowse insights index
AMD customer relationship map

AMD’s customer map: who drives revenue for the chipmaker and why it matters to investors

Advanced Micro Devices sells high‑performance CPUs and GPUs—EPYC server processors, Ryzen client CPUs, and Instinct AI accelerators—primarily to hyperscalers, OEMs/ODMs, distributors and enterprise customers. AMD monetizes through product sales, multi‑year capacity agreements with hyperscalers, and volume OEM channels, with international sales representing the majority of revenue and a mix of short‑term purchase orders and larger strategic contracts driving cash flow. For investors, the critical questions are revenue concentration, the degree of hyperscaler lock‑in, and how AMD’s contracting posture translates into predictable topline growth. Read more at https://nullexposure.com/ for deeper customer intelligence.

How AMD sells and the structural constraints that shape its revenue

AMD’s commercial model mixes spot and short‑term purchasing with a small number of very large, enterprise customers that drive outsized demand for data‑center products. AMD’s own FY2025 10‑K notes that sales are generally made under purchase orders or short‑term agreements and that no single customer exceeded 10% of revenue in 2025, even though a single client accounted for 18% in 2023—evidence of episodic concentration and shifting counterparty importance. International markets are material (67% of net revenue in 2025), and the company sells through both direct OEM channels and independent distributors that receive return and price‑protection rights. These facts imply a contracting posture that is largely short‑term/spot for most customers, anchored by a smaller set of very large hyperscaler and OEM relationships that are highly critical to AMD’s data‑center momentum.

  • Contracting posture: Predominantly short‑term and purchase‑order driven, with pockets of multiyear capacity deals (company 10‑K and earnings commentary, FY2025).
  • Customer concentration: Mixed — no 10% customers in 2025 but historical single‑customer concentration existed (FY2023–2025 filings).
  • Geography and reach: Global sales footprint with major revenue from outside the U.S. (FY2025 10‑K).
  • Channel mix: Direct OEM sales plus independent distributors with return/price protections (FY2025 filings).

Explore customer‑level signals and implications on the Null Exposure homepage: https://nullexposure.com/.

Who AMD is selling to right now — the complete relationship roll call

G42

In AMD’s 2025 Q3 earnings call AMD disclosed that Cisco and G42 will deploy a large‑scale AI cluster powered by Instinct MI350X GPUs, indicating G42 is a purchaser for hyperscale GPU deployments (AMD 2025 Q3 earnings call, cited Mar 2026).

Zyphra

AMD said IBM and Zyphra will train multiple generations of multimodal models on a large MI300X cluster, identifying Zyphra as a training‑workload customer for MI300X accelerators (AMD 2025 Q3 earnings call, Mar 2026).

OpenAI (earnings call)

AMD announced a comprehensive multiyear agreement with OpenAI to deploy 6 gigawatts of Instinct GPUs, establishing OpenAI as a major long‑term GPU capacity customer (AMD 2025 Q3 earnings call, Mar 2026).

Microsoft Corporation

AMD’s FY2025 10‑K warns that losing Microsoft’s support for software on AMD products would materially affect sales, underlining Microsoft as a critical ecosystem partner for AMD’s platforms (AMD 10‑K, FY2025).

AWS

Earnings commentary stated EPYC‑powered public cloud offerings grew significantly and that AWS, Google and others launched more than 230 new AMD instances during the quarter, signaling AWS as a fast‑growing EPYC cloud customer (AMD 2025 Q4 earnings call, Mar 2026).

Google (earnings call)

Google was named among hyperscalers launching new EPYC instances (over 230 total), confirming continued cloud demand for AMD CPUs (AMD 2025 Q4 earnings call, Mar 2026).

Tata Consultancy Services

AMD announced a strategic partnership with Tata Consultancy Services to co‑develop industry‑specific AI solutions and assist customers with AI deployments, identifying TCS as a systems and services partner for commercial AI rollouts (AMD 2025 Q4 earnings call, Mar 2026).

Meta Platforms (news)

Multiple news reports describe a multi‑year, gigawatt‑scale agreement with Meta to purchase in excess of $100 billion in AI computing capacity, positioning Meta as a transformational hyperscaler customer for Instinct GPUs and EPYC CPUs (news reports, FY2026).

Cisco

AMD disclosed that Cisco will deploy AI clusters using Instinct MI350X GPUs in partnership deployments such as the UAE project, marking Cisco as a systems integrator customer for GPU infrastructure (AMD 2025 Q3 earnings call, Mar 2026).

Oracle

Oracle publicly offered MI355X instances and was called out as the first hyperscaler to do so, indicating Oracle is provisioning AMD GPUs for real‑time inference and multimodal workloads (AMD 2025 Q3 earnings call, Mar 2026).

Alibaba

AMD noted hyperscalers, including Alibaba, launched new EPYC‑powered instances—evidence Alibaba is an EPYC cloud customer (AMD 2025 Q3 earnings call, Mar 2026).

Super Micro Computer

Press coverage highlighted Supermicro’s new MicroBlade platform featuring AMD EPYC 4005 processors, showing Supermicro as an OEM partner integrating EPYC into high‑density server products (news coverage, FY2026).

Microsoft Azure

AMD called out Microsoft Azure among hyperscalers launching new EPYC‑powered instances, confirming Azure as an EPYC cloud buyer (AMD 2025 Q3 earnings call, Mar 2026).

HPE

AMD stated it is partnering with DOE and Oak Ridge along with OCI and HPE to build Lux AI—positioning HPE as an infrastructure partner for science‑grade AI facilities (AMD 2025 Q3 earnings call, Mar 2026).

IBM

AMD noted IBM will train future multimodal models on MI300X clusters (with Zyphra), signaling IBM as a customer for large MI300X deployments (AMD 2025 Q3 earnings call, Mar 2026).

Lenovo

Press coverage described Ryzen AI 400 Series processors enabling OEMs including Lenovo to offer next‑gen AI PCs, showing Lenovo as a client‑PC OEM for AMD silicon (news coverage, FY2026).

HP

HP was listed alongside Lenovo as an OEM beneficiary of the Ryzen AI 400 Series, indicating HP is an AMD client CPU partner for AI‑enabled PCs (news coverage, FY2026).

Meta (news duplicate)

Additional press corroborated the same multi‑year Meta engagement and emphasized its revenue and credibility impact for AMD’s AI roadmap (news coverage, FY2026).

OpenAI (news duplicate)

Media reports echoed AMD’s strategic multibillion‑dollar collaboration with OpenAI and framed it as a record‑setting supply relationship supporting AMD’s AI positioning (news coverage, FY2026).

Meta Platforms (additional news)

Further outlets reiterated the transformational characterization of the Meta deal and the potential for AMD to add additional gigawatt‑scale customers (news coverage, FY2026).

Meta Platforms (another news item)

A separate news item reiterated the $100B+ capacity figure tied to the AMD–Meta partnership, emphasizing scale and investor perception (news coverage, FY2026).

OpenAI (news sentiment)

Market write‑ups flagged OpenAI and the possibility of AMD adding a third gigawatt‑scale customer alongside Meta and OpenAI, underscoring continued hyperscaler traction (news coverage, FY2026).

Meta (market commentary)

Analyst and investor commentary framed the Meta contract as validating Instinct GPUs and EPYC CPUs for hyperscaler AI builds, calling it a material multi‑year revenue opportunity (market commentary, FY2026).

Lenovo / HP (tradingview press)

Coverage on new Ryzen AI parts specifically named OEMs such as HP and Lenovo as launch partners to capture next‑gen AI PC demand (news coverage, FY2026).

OpenAI (finviz/trading sentiment)

Additional market stories repeated AMD’s multibillion collaboration with OpenAI and framed it as strategic to supply the leading AI company (news coverage, FY2026).

Investment implications: the bullish and the risk factors

Bull case drivers: Large multiyear hyperscaler contracts (Meta, OpenAI), broad hyperscaler EPYC adoption (AWS, Google, Azure, Oracle, Alibaba), and OEM wins (Supermicro, HP, Lenovo) create a diversified route to scale for both GPU and CPU revenue streams. Hyperscaler validation materially increases AMD’s addressable AI revenue over multiple years.

Key risks: Contracting remains largely short‑term and purchase‑order driven for many customers, which introduces topline volatility despite headline multiyear deals (AMD 10‑K, FY2025). A small number of customers still account for a disproportionate share of revenue historically, and distribution rights (returns/price protection) compress margins in channel sales. Concentration and spot purchasing are the principal execution risks.

If you want a structured view of these customer relationships and how they change quarter‑to‑quarter, see the full offering at https://nullexposure.com/.

What to watch next (practical signals)

  • Quarterly realizations of announced gigawatt‑scale capacity (bookings vs. shipments).
  • Hyperscaler instance rollouts and EPYC instance growth (AWS/Google/Azure cadence).
  • OEM product launches and design‑wins (Supermicro, HPE, Lenovo, HP).
  • Any changes to distributor return/price protection policies or customer concentration disclosures in the next 10‑K/10‑Q.

For ongoing customer signals and to monitor how these relationships convert into revenue, visit https://nullexposure.com/ — the research portal built for monitoring enterprise vendor exposure.