AMD’s Data Center AI business, its Instinct GPUs and Helios racks, will be larger than its server CPU business for the first time in the fourth quarter, Lisa Su said on the company’s earnings call on 3 November: “Q4 is the first quarter where the data center AI business will cross over in terms of size compared to the CPU business”.
AMD does not report the two businesses separately. But its CFO guided Data Center revenue up about 50% in the fourth quarter, to about $12.8 billion by our arithmetic, so a larger half puts Data Center AI above about $6.4 billion. In the third quarter, when server CPUs were still the larger business, Data Center AI was below about $4.27 billion.
Business Pill · The Memory Wall
An evergreen Business Engineer explainer on the memory wall in AI chips. It teaches the general idea only and does not discuss AMD’s quarter.
The key insight: By our arithmetic from Hu’s guide, AMD’s Data Center AI revenue goes from below about $4.27 billion in the third quarter to above about $6.4 billion in the fourth, at least 1.5 times, in the quarter it passes server CPUs.
Chief financial officer Jean Hu raised the crossover first, about 22:41 UTC, answering a TD Cowen analyst’s question on gross margin: “And in Q4, we are guiding sequentially revenue growing 50% for data center business. It’s actually going to be the quarter data center AI business surpass server business.” She added that the AI business is “a little bit of dilutive to corporate average, but gross margin dollars are increasing very significantly”.
Su confirmed it about 22:44 UTC, in answer to a UBS analyst asking how Data Center would split in 2027. She said: “For the data center CPU business, we will about double, and for the data center AI business, it will significantly more than double.”
Earlier in the Q&A she said “The third quarter was the first quarter that we shipped MI455 and Helios racks for revenue”, and added: “We have a big step up in Q4. We’ll continue to ramp throughout 2027.”
AMD’s Data Center segment revenue in the third quarter of 2026 (reported, 8-K Exhibit 99.1) and the fourth quarter implied by CFO Jean Hu’s approximate 50% sequential guide on the 3 November call; the dashed line marks half the segment, which bounds Data Center AI once Su and Hu say AI passes server CPUs in the fourth quarter. The implied figure and bounds are our arithmetic.
Who Is Running Helios
In her prepared remarks Su said: “We started shipping MI455X in volume in the quarter, and lead customers including Anthropic, Meta, Microsoft, OpenAI, and Oracle brought their first Helios racks online quickly.” She said AMD expects “Helios deployments to increase substantially in the fourth quarter” as HPE, Dell, Lenovo, Supermicro and other partners begin shipping systems.
A Bank of America analyst asked about AMD’s confidence that anchor customers have the financing and sites for what he called “roughly 3 gigawatt or so of planned AMD deployments into 27”. Su answered: “We have been planning for this ramp with our customers jointly. So we have good visibility into the 2027, as you said, land, power, shells, all of those, the data center infrastructure.”
She added that “we know where the GPUs are going in terms of which data centers and which locations”, and that AMD is “selectively investing with our lead partners” where it helps to have the data center infrastructure identified in advance.
The Arithmetic
Data Center revenue was $8.546 billion in the third quarter. A rise of about 50%, as Hu guided, gives about $12.8 billion in the fourth quarter, by our arithmetic. If AI is the larger of the two businesses in that quarter, it is above about $6.4 billion; if server CPUs were the larger in the third quarter, AI was below about $4.27 billion then.
Together, those bounds put fourth-quarter Data Center AI revenue at least about 1.5 times its third-quarter level, by our arithmetic. The Data Center segment also includes other products, such as networking cards and FPGAs, so the true split could differ; AMD gave no dollar figure for either business.
Update, 3 November, later in the Q&A
Asked by a Bernstein analyst how large the crossover is, about 22:49 UTC, Hu said “the data center AI will be much higher than server business in Q4”, and confirmed she meant “in terms of dollars”. The above-$6.4 billion figure in this piece is our arithmetic and assumed only that AI is the larger of the two businesses, half of about $12.8 billion; Hu gave no dollar figure.
Quotations from our own machine transcript of the call audio; times approximate.
How Each Side Reads These Figures
For readers who see a ramp, the statements include five named labs and clouds running Helios, a first revenue quarter for MI455 and Helios in the third quarter, and Su’s statement that the AI business will “significantly more than double” in 2027.
For readers who see risk, Hu said the AI business dilutes gross margin against the corporate average, and the analyst’s 3 gigawatts of 2027 deployments depend on customers’ land, power and financing, which Su addressed only in general terms. We cite no analysts’ views beyond their questions.
As we read it, the crossover gives the first public bound on a number AMD does not report: how big its GPU business is.
As we read it, the margin trade-off is explicit: Hu said the AI business dilutes gross margin against the corporate average while adding gross margin dollars.
As we read it, Su answered the 3 gigawatt question with planning detail, sites and locations, rather than a dollar or gigawatt figure of her own.
Lisa Su, AMD earnings call, 3 November 2026 (our transcript)
“Q4 is the first quarter where the data center AI business will cross over in terms of size compared to the CPU business”
Three Implications
FOR NVIDIA WATCHERS Su named Anthropic, Meta, Microsoft, OpenAI and Oracle as lead customers that brought their first Helios racks online in the third quarter.
FOR AMD SHAREHOLDERS Hu said the AI business is a little dilutive to gross margin against the corporate average, while gross margin dollars rise.
FOR DATA CENTER BUILDERS Su said AMD knows which data centers and locations its GPUs are going to in 2027 and is selectively investing with lead partners.
The framework puts it this way: “Inference is a revenue engine: continuous, distributed across millions of applications and agents, incurring marginal cost with every query.”
As we read it, Su tied fourth-quarter growth to increased demand for inference and agentic workloads. The framework was published in March 2026 and does not discuss AMD’s fourth quarter.
Quotations come from our own machine transcript of the call audio, as rebroadcast live; it is not an official AMD transcript and times are approximate. AMD does not report Data Center AI and server CPU revenue separately, so the $6.4 billion and $4.27 billion bounds and the 1.5 multiple are our arithmetic, based on Hu’s approximate 50% Data Center guide and the segment’s $8.546 billion third-quarter revenue.
The 3 gigawatt figure is the Bank of America analyst’s description, not an AMD disclosure. We did not contact AMD.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Su and Hu said AMD’s AI GPUs will outsell its server CPUs for the first time in the fourth quarter. By our arithmetic from Hu’s guide, that puts fourth-quarter Data Center AI revenue above about $6.4 billion, against less than about $4.27 billion in the third quarter, the first quarter in which MI455 and Helios racks shipped for revenue.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. Quotations from Lisa Su and Jean Hu come from our own machine transcript (faster-whisper) of AMD’s third-quarter 2026 earnings call of 3 November, as rebroadcast live; it is not an official transcript and times are approximate. The Data Center figure comes from AMD’s release filed as Exhibit 99.1 to its 8-K of 3 November. The fourth-quarter estimate and the bounds are our arithmetic. We did not contact AMD.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
