SpaceX首份盈利季报大超预期，云业务推动并给出2027年最高13GW算力与5000亿至1万亿美元ARR愿景

SpaceX ( SPCX ) CEO Elon Musk outlined a massive amount of compute capacity his company plans to bring online over the next year and how demand for that capacity continues to outpace supply. The Memphis Sumpercompute cluster Colossus II is bringing about 200,000 Nvidia ( NVDA ) GB300s online per month, which will continue throughout the end of the year, he said during Tuesday night's earnings call . That pace is expected to continue into 2027, with the new compute primarily coming from Nvidia's new Vera Rubin processors. "At this point, I would say—I would say I'm fairly confident of being able to bring an incremental 10GW of compute online in 2027," Musk said. "That would be in addition to the roughly 3GW, maybe 2.5 to 3GW, that we'll have online by the end of this year … I would be surprised if we brought on less than 13GW of compute by the end of next year." In the same time frame, Musk said they plan to have 20GW of power and cooling available to have plenty of excess infrastructure needed to run the GPUs. By the end of the second quarter of fiscal 2027, SpaceX plans to have 440,000 Vera Rubins deployed for optimal training. "This will, I think, be extremely helpful for getting Grok to be potentially the best model, but certainly to be extremely competitive," Musk said. "So on the Grok bot side, we're seeing the fastest growth of a product that I've ever seen in my life." Paid subscribers of Grok Bot, an agentic assistant, more than doubled month over month to 800,000. "So I mean, in a nutshell, what this implies for SpaceX's annual recurring revenue by the end of '27 is somewhere between $500B and $1T," Musk added. Cloud computing appears to be driving substantial revenue gains for SpaceX. "Segment revenue increased substantially to $7.6B, up $5B quarter-over-quarter and $6.7B year-over-year, primarily powered by the ramp of our cloud services agreements," said SpaceX Chief Financial Officer Bret Johnsen. "There are many customers asking for compute, and we're actively building the supply to meet both internal and external demand," Johnsen added. "We expect to have additional new customers in future deals, and we'll continuously evaluate the highest margin usage for our compute assets across both new and existing cloud service agreements and serving inference for our own SI products." "At current pricing, new compute deployments are generating a payback of less than a year," Johnsen said. "Pricing is currently tracking slightly above the high end of the $30 to $50 per watt range we were discussing just last quarter, which further supports the attractiveness of these terrestrial SI compute investments." Total company capital expenditures during the third quarter were $19.6B. Because of the additional capacity coming online this quarter, SpaceX expects fourth-quarter capex to range from $35B to $40B. Cloud computing prices continue to rise Regarding compute usage, Musk said the goal of internal use and external allocation to third parties would ideally run at a 50-50 split. However, that can change from month to month. "Our belief is that the value of compute per watt will increase over time, not decrease," Musk said. "And the reason for that, I think it should be obvious to people, is that as you improve the AI software, as the models get better and better … you get more useful intelligence per watt." "And that's why you see things like even old computers like H100s renting for more than they rented when they were new," Musk added. "And I think that's likely to continue." Musk expects a shortage in AI compute to linger for a "long time." "We are also, frankly, hounded relentlessly by every hyperscaler and by dozens of smaller companies to rent more of our compute," he said. "So we could rent out our compute several times over if we said yes to everyone. That's just the situation here." Terafab progress Regarding progress on the Texas foundry Terafab, Musk hopes to begin testing chip functionality by May or June of 2027. "That's just a kind of SRAM chip, something that basically tests the line," he said. "That's at the research fab in Austin. Our aspiration is to reach volume production around Q3 of '28 at Terafab at Gibbons Creek, the main Terafab location." "Now these are very aggressive dates, and we're very appreciative of ASML ( ASML ) and all of our other suppliers in accelerating equipment deliveries in order to achieve this outcome," Musk added.