【TMGM Financial Recap】Two Earnings Reports: AI Infrastructure Concerns Recede Temporarily as AMD and Supermicro Soar After Hours

AMD initially gained more than 9%. The advance eased to around 6.85% in trading after the close on 3 November, while Supermicro was up nearly 10% at one stage. Both stocks rose as their earnings reports, released that evening, delivered a shared message from different points in the AI infrastructure supply chain. AMD supplies the computing chips; Supermicro incorporates them into servers and delivers those systems to data centres.

The previous week’s losses give that message particular weight. Alphabet’s disclosure of a second consecutive quarter of negative free cash flow, alongside its forecast for a “significant increase” in capital expenditure for 2027, drove a 10% after-hours fall on 28 October. The stock finished the following day 13% lower. Intel lost 30% in one session on 30 October, following fourth-quarter guidance well short of expectations and an indication that full-year free cash flow would remain negative. These developments are shaping how the market prices AI: capital expenditure continues to expand, yet the timing of returns looks increasingly uncertain. Rather than directly establishing when returns will emerge, AMD and Supermicro addressed the underlying question of whether AI infrastructure orders remain in place.

97% Data Centre Growth: AMD Guidance Outweighs Earnings Beat

Across all measures, AMD’s third-quarter earnings cleared market expectations. Revenue of $13.377 billion represented 45% annual growth, exceeding the $13.09 billion forecast. Against an expected $1.95, adjusted earnings per share stood at $2.09. Gross margin reached 56%, an increase of two percentage points year on year.

Data centres accounted for the main impetus behind the after-hours move. The division generated $8.546 billion in revenue, a 97% annual increase. Its contribution to total revenue consequently reached 64%, compared with 47% a year earlier. Two factors supported that performance: robust EPYC server CPU demand and rising output of Instinct MI450 series GPUs. As AMD’s AI accelerator product line competing with Nvidia, the MI450 reported “record sales”. That suggests AMD now has a quantifiable position in AI computing, rather than merely a product roadmap.

The outlook carried greater weight than the reported quarter. For the fourth quarter, AMD projects approximately $17 billion in revenue, plus or minus 3%. Relative to the average analyst estimate of approximately $16.3 billion, the midpoint is around $700 million higher. It implies around 27% growth quarter on quarter and approximately 66% year on year. Management therefore expects growth to gather further pace after the third quarter’s near-doubling. In her earnings-call remarks, Lisa Su highlighted capacity expansion for Helios rack-scale AI systems and sixth-generation EPYC “Venice” processors. She stated explicitly that “data centre AI will grow faster than the company previously expected.”

The strength was not uniform across the business. At $3.722 billion, combined client and gaming revenue was 8% below the year-earlier level; gaming alone declined 40%. Inventories, accounts receivable and capital expenditure all rose, while free cash flow contracted approximately 64% year on year to $554 million. The market had expected around $1.05 billion.

Nearly 300% Revenue Growth: Supermicro Margin Rebound Offers Reassurance

Supermicro generated $19.9 billion in first-quarter fiscal 2027 revenue. That was approximately 298% higher year on year and substantially above the $16.4 billion market estimate. Adjusted earnings per share of $1.81 beat the expected $1.23 by approximately 47%. Management set its second-quarter revenue range at $19.5 billion to $21.5 billion, against expectations of approximately $17.1 billion to $18 billion. For the full year, it upgraded revenue guidance to $72 billion–$82 billion from $65 billion–$72 billion.

AI server integration and delivery sit at the centre of Supermicro’s business. Revenue therefore directly tracks the rate at which cloud providers, AI laboratories and enterprises deploy computing capacity within data centres. It does not measure orders at the chip design stage. The 298% annual expansion indicates that physical AI infrastructure construction remains rapid.

Profitability was an even more important feature of the release. On a GAAP basis, first-quarter gross margin reached 12.7%; the non-GAAP figure was 12.8%. Both exceeded the year-earlier levels of 9.3%–9.5% and analysts’ previously pessimistic estimates. For investors, revenue expansion was no longer the only issue. Their central concern was Supermicro’s ability to preserve profitability as Nvidia chip supplies remained tight and memory prices rose. Recovering gross margin provided partial reassurance.

That reassurance should not be mistaken for the removal of other risks. Recent structural pressure on Supermicro has come from legal issues as well as changes in the AI cycle. News that the US Department of Justice had brought export-related charges against three individuals, including a co-founder, previously sent the shares down 33% the following day. The latest after-hours earnings rally partly reflects financial fundamentals temporarily eclipsing legal concerns, not legal risk disappearing. The gain also does not yet fully reflect the company’s CFO change or the operating constraints imposed by GPU and processor shortages.

298% Server Growth: Faster AI Deployment, Uneven Benefits

Chip orders are gaining momentum, as AMD’s data centre revenue demonstrates. The MI450 production ramp-up also shows that buyers are absorbing computing capacity offered by suppliers beyond Nvidia. At the server level, Supermicro’s revenue and forecasts signal that deployment has maintained its pace. Its 298% growth reflects continuing expansion of physical data centre capacity. These reports together support a specific assessment: at least for infrastructure orders and deployment, there is no evidence of weakening fundamentals to substantiate the “AI capital expenditure anxiety” triggered by Intel and Alphabet in late October.

This assessment has a clear boundary: healthy orders and deployment do not establish that returns have been realised. AMD continues to report declining free cash flow. Supermicro’s margin recovery still leaves margins around 12%, and Alphabet and Amazon continue to face free cash flow shortfalls. The pattern across the AI infrastructure chain is uneven. Figures look stronger for businesses closer to construction, including chips and servers. Pressure is more evident closer to monetisation, measured through cloud services’ free cash flow and advertising conversion efficiency.

For now, the gains in AMD and Supermicro after hours demonstrate that investors will pay a premium for orders that continue to gather pace. Once regular trading begins, sustaining that premium and extending it throughout the AI infrastructure chain will depend on one judgement: whether investors regard order strength as sufficient to offset uncertainty over when returns will arrive.