
After the close on November 3, shares of AMD and Supermicro moved in the same direction. AMD initially surged more than 9% before trimming its gain to around 6.85%, while Supermicro climbed nearly 10% at one point. The two companies occupy different positions in the AI infrastructure supply chain: AMD supplies computing chips, while Supermicro assembles those chips into servers for delivery to data centres. Their earnings reports pointed to the same conclusion on the same evening.
That signal carries greater significance against the backdrop of the previous week. On October 28, Alphabet fell 10% after hours after reporting negative free cash flow for a second consecutive quarter and forecasting a “significant increase” in capital expenditure for 2027. It closed 13% lower the following day. On October 30, Intel plunged 30% in a single session after issuing fourth-quarter guidance well below expectations and indicating that full-year free cash flow would remain negative. A pricing framework is taking shape in the market: AI capital expenditure is expanding, but the timeline for returns is becoming less clear. AMD and Supermicro did not directly answer the question of when those returns would materialise, but they answered a more fundamental question: are AI infrastructure orders still there?

AMD: Data Centre Revenue Nearly Doubles, but Guidance Matters More Than the Earnings Beat
AMD’s third-quarter results exceeded expectations across the board. Revenue reached $13.377 billion, up 45% year on year and above market expectations of $13.09 billion. Adjusted earnings per share came in at $2.09, versus expectations of $1.95, while gross margin rose two percentage points year on year to 56%.
The data centre business was the real driver of the after-hours reaction. Revenue in the segment reached $8.546 billion, up 97% year on year, with its share of total revenue rising from 47% a year earlier to 64%. Growth came from two sources: strong demand for EPYC server CPUs and the ramp-up of Instinct MI450 series GPUs. The MI450 is AMD’s AI accelerator product line competing with Nvidia. Its “record sales” suggest that AMD has established a measurable foothold in the AI computing market, moving beyond the product roadmap stage.
Fourth-quarter guidance mattered even more than the third-quarter figures. AMD expects revenue of approximately $17 billion, plus or minus 3%. The midpoint is around $700 million above analysts’ average estimate of approximately $16.3 billion, implying sequential growth of around 27% and year-on-year growth of approximately 66%. The significance of this guidance is that, following the near-doubling recorded in the third quarter, management expects growth to accelerate further. On the earnings call, Lisa Su cited expanding production capacity for Helios rack-scale AI systems and sixth-generation EPYC “Venice” processors, explicitly stating that “data centre AI will grow faster than the company previously expected.”
However, the report also revealed areas of divergence. Combined client and gaming revenue fell 8% year on year to $3.722 billion, with gaming revenue declining 40%. Free cash flow dropped approximately 64% year on year to $554 million, below market expectations of around $1.05 billion, as inventories, accounts receivable and capital expenditure all increased.
Supermicro: Revenue Growth of Nearly 300% and Margin Recovery Bring Relief
Revenue for the first quarter of fiscal 2027 reached $19.9 billion, up approximately 298% year on year and well above market expectations of $16.4 billion. Adjusted earnings per share came in at $1.81, approximately 47% above expectations of $1.23. Second-quarter revenue guidance was set at $19.5 billion to $21.5 billion, compared with market expectations of approximately $17.1 billion to $18 billion. Full-year revenue guidance was raised from $65 billion–$72 billion to $72 billion–$82 billion.
Supermicro’s core business is integrating and delivering AI servers. Its revenue growth directly reflects the pace at which cloud providers, AI laboratories and enterprises are deploying computing capacity in data centres, rather than orders at the chip design stage. Year-on-year growth of 298% indicates that the physical construction of AI infrastructure is still moving at a rapid pace.
An even more notable detail was its margins. Supermicro reported a first-quarter GAAP gross margin of 12.7% and a non-GAAP gross margin of 12.8%, above the 9.3%–9.5% levels recorded a year earlier and analysts’ previously pessimistic estimates. The market’s central concern about Supermicro had extended beyond revenue growth to whether it could protect profitability amid tight Nvidia chip supplies and rising memory prices. The recovery in gross margin partly eased that concern.
However, the nature of this relief needs to be clearly understood. Supermicro has recently faced more than fluctuations in the AI cycle; it has also been under structural pressure from legal issues. Export-related charges brought by the US Department of Justice against three individuals, including a co-founder, previously triggered a 33% share-price plunge the day after the news emerged. Part of the after-hours rally following these earnings reflects fundamentals temporarily overshadowing legal concerns, rather than the disappearance of legal risk. The change in the company’s CFO and business constraints caused by GPU and processor shortages also remain factors that the after-hours gain has yet to fully account for.
AI Deployment Is Still Accelerating, but the Benefits Are Uneven
AMD’s data centre revenue demonstrates that chip orders are accelerating, while the ramp-up of the MI450 indicates that the market is absorbing computing capacity supplied by companies beyond Nvidia. Supermicro’s revenue and guidance show that server deployment has not slowed, with its 298% growth reflecting the continued expansion of physical data centre capacity. Together, these results support one assessment: the “AI capital expenditure anxiety” triggered by Intel and Alphabet in late October has, at least at the level of infrastructure orders and deployment, found no evidence of deteriorating fundamentals.
However, that assessment comes with a qualification: strong orders and deployment do not mean returns have materialised. AMD’s free cash flow is declining. Although Supermicro’s margins have recovered, they remain around 12%, while the free cash flow shortfalls at Alphabet and Amazon persist. Along the AI infrastructure chain, the closer a business is to the construction end—chips and servers—the stronger its figures appear. The closer it is to monetisation—free cash flow from cloud services and advertising conversion efficiency—the more visible the pressure becomes.
The after-hours rallies in AMD and Supermicro show that, at this point, the market is willing to pay a premium for orders that are still accelerating. Whether that premium can hold after the market opens, and whether it can spread across the entire AI infrastructure chain, depends on whether investors believe strong orders are enough to offset doubts about the timeline for returns.