AMD's Second Route Into the AI Data Center Runs Through Other Companies' Chips

Advanced Micro Devices has spent the past two years being measured almost entirely against Nvidia's accelerator roadmap. A pair of developments this week suggests investors should widen the lens. The chipmaker now sits inside AI systems whose headline compute comes from someone else — and that position may matter as much as the race to sell Instinct cards outright.

Google Cloud's TPU 8tx runs on AMD's fifth-generation EPYC processors, which handle orchestration, data preparation and communication with the TPUs. Put simply, AMD collects a slice of the budget even when the flagship silicon is Google's. If such mixed-architecture builds become the norm, the company can ride the diversification of AI infrastructure without having to win every accelerator order with its own hardware.

That dual role is the heart of the bull case. Server processors open a second door into customers' capital-spending plans, running alongside the Instinct accelerator line. The catch is that the same structural shift cuts both ways. William Blair flags custom chips from Meta and OpenAI as a competitive threat to Instinct — while noting that CPU sales into those very systems could work in AMD's favor. Whether the trade-off nets out positive depends on whether extra processor business can economically offset any accelerator share lost.

An Upgrade Built on a Rival's Stumble

The more immediate catalyst arrived yesterday, when William Blair lifted AMD from "Market Perform" to "Outperform." Among the reasons cited: delays and supply bottlenecks afflicting Intel's Xeon processors. A constrained competitor is only worth something if AMD can actually serve the demand that spills over.

The analysts see more than 80% revenue growth as possible for 2027, and expect AMD's data-center AI revenue to overtake its server-CPU business in the fourth quarter of 2026 — a meaningful shift in the company's internal balance. Management's own guidance puts fourth-quarter 2026 revenue at roughly $17 billion, give or take 3%, which turns the broad growth narrative into a checkable target.

The base is already solid. AMD reported third-quarter 2026 revenue of $13.377 billion on November 3, up 45% year on year. Data-center revenue climbed 97% over the same period, a gain the company attributed to EPYC demand and the production ramp of MI450.

Quantum, Open-Source Software and a 6.1% Pop

The stock gained 6.1% yesterday, helped along by the upgrade and a friendlier backdrop for Nvidia's challengers, as media coverage framed the move. Two fresh initiatives broaden the story further. Reuters reported that AMD plans an equity investment in PsiQuantum as part of an expanded partnership to jointly develop quantum-computing infrastructure. Separately, LMSYS Org launched the SGLang Consortium with AMD as a founding member, aimed at advancing open-source software for AI inference.

The two ventures pull in different directions — inference software on one side, quantum hardware on the other — and neither will show up in the top line soon. Their value to shareholders hinges on whether the technical collaboration translates into wider adoption of AMD technology.

The Financing Question Hanging Over Everything

The most important caveat concerns how the entire buildout gets paid for. Reuters reported investor concerns that Nvidia may finance fewer AI data centers following Jensen Huang's departure. TrendForce, meanwhile, sees room for alternative suppliers to grab market share faster amid the leadership change. For AMD, that short-term tailwind lines up with the longer structural trend — but its economic worth is measured in systems shipped and revenue booked, not in sentiment.

A concrete test comes with the promised general availability of cloud instances running sixth-generation EPYC processors at Microsoft Azure and Oracle Cloud Infrastructure, which AMD has said are coming soon. Broader availability could open additional use cases, provided it converts into customer demand.

Weighing it all up, the convincing case for AMD rests on earning across multiple architectures — holding its server-processor position while building Instinct into a genuine second engine. The partnerships and design wins widen the runway. What investors should watch is whether that technical breadth turns into durable additional business, because that is where the expected growth spurt will either be confirmed or found wanting.