AMD's data-centre revenue nearly doubled in Q3. Plan AI infrastructure for tight supply

In brief

AMD reported third-quarter 2026 results on 3 November (US time). Revenue was a record $13.4 billion ($13,377 million), up 45% year on year. Data Center segment revenue was a record $8.5 billion, up 97%, which AMD attributes to EPYC server processors and the launch of its Instinct MI450 Series GPUs. CNBC reported that revenue beat the LSEG consensus of $13.09 billion.

For the fourth quarter, AMD expects revenue of approximately $17 billion, plus or minus 3%, which it says would be about 66% growth year on year. It also says Anthropic, Meta, Microsoft, OpenAI and Oracle Cloud Infrastructure are running models and workloads on the first AMD Helios rack-scale systems.

What the numbers say about AI demand

Data Center now accounts for roughly 64% of AMD's revenue, against 47% a year earlier, based on the segment figures in AMD's release. Chair and CEO Lisa Su said AMD expects "growth to accelerate further in the fourth quarter as we scale Helios and ‘Venice’ production."

Intel's results on 29 October point the same way on servers. Intel reported Data Center and AI revenue of $7.3 billion, and CFO David Zinsner cited "our strongest data center revenue growth in 15 years." But Intel guided fourth-quarter revenue to $15.0 billion to $16.0 billion, below the $17.1 billion analysts expected according to CNBC, and Zinsner said: "we see the PC slowdown continuing at least through the first half of next year." On its earnings call Intel also said it expects the industry to remain supply-constrained through 2027 and into 2028.

Read together, the two companies describe the same market: spending is concentrated in data-centre CPUs and accelerators, while PCs weaken. That is ByteNib's reading of two vendors' own reports, not independent demand data, and guidance is not a delivery commitment.

What IT and infrastructure buyers should take from it

1. Expect allocation, not discounts. When the largest AI labs and hyperscalers are taking the first rack-scale systems, enterprise orders for current-generation accelerators and high-core-count servers can end up queuing behind them. Confirm lead times in writing before committing project dates.
2. Price the whole rack. Helios is sold as rack-scale infrastructure. Check power density, liquid cooling and floor loading with your facility or colocation provider before a pilot, not after.
3. Keep a second source. AMD's ROCm software and Intel's server roadmap both give buyers leverage against a single accelerator supplier. Test your main models on more than one stack while you can still choose.
4. Separate PC and server budgets. Intel's PC warning suggests client devices may cost more or arrive later. Do not let a delayed laptop refresh quietly absorb money set aside for infrastructure, or the reverse.

Sources and scope

Financial figures, the Helios customer list and the Lisa Su quote are from AMD's third-quarter 2026 results release of 3 November 2026. The revenue share calculation uses that release's segment table. Intel's guidance, Data Center and AI revenue and the Zinsner quotes are from Intel's third-quarter 2026 results release of 29 October 2026; the supply outlook is from Intel's earnings call as transcribed by The Motley Fool. Consensus estimates are from CNBC's AMD report and CNBC's Intel report, citing LSEG. All checked on 4 November 2026. Guidance is forward-looking and vendor-stated. The buyer recommendations are ByteNib editorial analysis.

Drafted with AI assistance and reviewed by the editor before publication.

Continue exploring: Cloud & DevOps analysis, the related implementation tutorial, and the structured learning path.