Shares of Super Micro Computer rocketed roughly 20% in after-hours trading on Tuesday. The artificial intelligence server maker released preliminary fiscal fourth-quarter results, and the numbers show a massive leap in profitability.
GAAP and non-GAAP gross margins are now expected to land between 15% and 17%. That is double the company’s prior guidance range of 8.2% to 8.4%. Management attributed the profitability spike to a favorable customer and product mix, driven by macro-level investments in AI infrastructure as cloud providers rapidly build out physical data centers.
But this was not just a margin story. The stock’s extended-trading rally was also fueled by a $60 billion order backlog accumulated during the fourth quarter alone. This surge in demand completely overshadowed a slight miss on the top line. Q4 revenue is now estimated to hit near the low end of the previously guided $11 billion to $12.5 billion range.
The Q4 turnaround is a sharp reversal in investor sentiment. Just last month, Super Micro’s stock nearly halved following the announcement of a $7 billion equity financing plan that sparked widespread fears of shareholder dilution. Earlier in the fiscal year, gross margins had dropped to a concerning 6.3%.
During the May earnings call, CEO Charles Liang promised a return to double-digit margins by prioritizing the high-margin Data Center Building Block Solutions enterprise business. That strategy is now generating concrete results, including an active project co-building a new data center alongside SpaceX.
SpaceX Partnership and Enterprise Pivot Reshape AI Hardware Valuations
Super Micro’s margin expansion immediately lifted the broader AI hardware sector in extended trading. Shares of major server competitors like Dell and Hewlett Packard Enterprise rallied simultaneously on Tuesday night.
By shifting focus away from aggressive market-share expansion and locking down premium enterprise contracts, Super Micro is proving that the secondary AI infrastructure market can actually sustain high profitability. The massive $60 billion backlog indicates that tech companies are still heavily scaling their physical operations to support large language models, dismissing recent analyst fears of an AI hardware spending slowdown.