【TMGM Financial Recap】AMD and Supermicro Rally After Hours: Have Two Earnings Reports Temporarily Eased AI Infrastructure Concerns?

AI infrastructure shares found support after the close on 3 November, with AMD and Supermicro both advancing. An initial surge of more than 9% in AMD narrowed to around 6.85%, while Supermicro rose nearly 10% at one stage. Although they serve different parts of the AI infrastructure supply chain, their earnings reports delivered a common message that evening. AMD provides computing chips; Supermicro builds those chips into servers for delivery to data centres.

The previous week’s sell-offs make that message more significant. After reporting negative free cash flow for a second consecutive quarter and projecting a “significant increase” in capital expenditure for 2027, Alphabet dropped 10% in after-hours trading on 28 October. The following day, it closed 13% lower. Intel then slumped 30% in a single session on 30 October, having provided fourth-quarter guidance far below expectations and signalled that full-year free cash flow would stay negative. Investors are developing a pricing framework in which rising AI capital expenditure sits alongside growing uncertainty over the timing of returns. Neither AMD nor Supermicro directly resolved that uncertainty. Instead, their reports addressed a more basic issue: whether demand for AI infrastructure orders remains intact.

AMD: Why Does Guidance Matter More Than an Earnings Beat and Near-Doubling of Data Centre Revenue?

Across its third-quarter results, AMD beat market forecasts. At $13.377 billion, revenue increased 45% year on year and surpassed expectations of $13.09 billion. Gross margin expanded by two percentage points from a year earlier to 56%, while adjusted earnings per share of $2.09 exceeded the expected $1.95.

Behind the after-hours share-price response was the strength of the data centre division. Its revenue climbed 97% year on year to $8.546 billion, taking its contribution to total revenue from 47% a year earlier to 64%. Robust demand for EPYC server CPUs and the scaling up of Instinct MI450 series GPUs underpinned that expansion. AMD competes with Nvidia through its MI450 AI accelerator product line. The line’s “record sales” indicate a measurable presence in AI computing, rather than a business still confined to a product roadmap.

More consequential than the third-quarter performance was the outlook for the fourth quarter. Revenue is expected by AMD to reach approximately $17 billion, plus or minus 3%. At the midpoint, that is around $700 million ahead of analysts’ average forecast of approximately $16.3 billion and implies growth of around 27% sequentially and approximately 66% year on year. In other words, management anticipates a further acceleration after revenue nearly doubled in the third quarter. During the earnings call, Lisa Su pointed to increased production capacity for Helios rack-scale AI systems and sixth-generation EPYC “Venice” processors. She explicitly stated that “data centre AI will grow faster than the company previously expected.”

Not all parts of the report showed the same strength, however. A 40% fall in gaming revenue accompanied an 8% year-on-year decline in combined client and gaming revenue to $3.722 billion. With inventories, accounts receivable and capital expenditure all rising, free cash flow fell approximately 64% from a year earlier to $554 million, missing market expectations of around $1.05 billion.

Supermicro: Why Are Nearly 300% Revenue Growth and Recovering Margins Easing Concerns?

For the first quarter of fiscal 2027, Supermicro delivered revenue of $19.9 billion, comfortably exceeding the market forecast of $16.4 billion and marking approximately 298% year-on-year growth. At $1.81, adjusted earnings per share were approximately 47% higher than expectations of $1.23. Against market forecasts of approximately $17.1 billion to $18 billion, the company projected second-quarter revenue of $19.5 billion to $21.5 billion. It also increased its full-year revenue outlook to $72 billion–$82 billion from $65 billion–$72 billion.

Integrating and delivering AI servers forms the core of Supermicro’s operations. Rather than capturing orders at the chip design stage, its revenue tracks how quickly cloud providers, AI laboratories and enterprises install computing capacity in data centres. The 298% year-on-year increase therefore points to a continued rapid pace of physical AI infrastructure construction.

Margins provided an even more significant detail. First-quarter gross margin stood at 12.7% on a GAAP basis and 12.8% on a non-GAAP basis, exceeding both the 9.3%–9.5% levels a year earlier and analysts’ previously pessimistic forecasts. Investors’ principal concern had moved beyond Supermicro’s ability to grow revenue: could it sustain profitability while Nvidia chips remained in tight supply and memory prices increased? The gross margin rebound offered some reassurance on that front.

That reassurance nevertheless needs to be viewed in context. Alongside fluctuations in the AI cycle, legal issues have recently placed Supermicro under structural pressure. After news of export-related charges brought by the US Department of Justice against three individuals, including a co-founder, its shares plunged 33% the next day. The post-earnings after-hours rally thus partly represents fundamentals temporarily taking precedence over legal concerns, not an end to legal risk. Nor has the after-hours advance fully accounted for the change in the company’s CFO or the operating constraints arising from shortages of GPUs and processors.

Why Are the Benefits Uneven as AI Deployment Continues to Accelerate?

Accelerating chip orders are evident in AMD’s data centre revenue, while the scaling up of the MI450 shows that buyers are taking up computing capacity from suppliers other than Nvidia. At Supermicro, revenue and guidance indicate no slowdown in server deployment; growth of 298% captures the ongoing addition of physical data centre capacity. Read together, the reports support an assessment that, at least in infrastructure orders and deployment, there is no evidence of weaker fundamentals behind the “AI capital expenditure anxiety” sparked by Intel and Alphabet in late October.

There is an important limit to that assessment: robust orders and deployment do not establish that returns have arrived. Free cash flow at AMD is falling, and despite their recovery, Supermicro’s margins are still around 12%. Alphabet and Amazon, meanwhile, continue to face free cash flow shortfalls. Businesses nearer the construction side of the AI infrastructure chain, namely chips and servers, are reporting stronger figures. Pressure becomes more apparent nearer the monetisation side, in free cash flow from cloud services and advertising conversion efficiency.

For now, the after-hours advances in AMD and Supermicro indicate investors’ willingness to attach a premium to orders that continue to accelerate. Its survival after the market opens, and its potential extension across the wider AI infrastructure chain, hinge on whether investors consider strong orders sufficient to counter uncertainty about when returns will emerge.