SpaceX、Anthropic、OpenAIのIPOが市場全体の40%暴落を引き起こす可能性は低いが、AI・テクノロジー株のローテーションを誘発する見込み

SpaceX is reportedly preparing to issue $40 billion in new debt to buy Nvidia chips for its data centers, according to Bloomberg, sending shares down 2% in early trade Wednesday. The financing would be split into $10 billion in bank loans and $30 billion in investment-grade debt, with Apollo Global leading the financing and bond giant PIMCO looking at the deal. SpaceX's data centers in Memphis, Tennessee, known as Colossus 1 and 2, are generating billions per month in recurring revenue from AI compute services, and CFO Bret Johnsen said last month the company believes it is on track to hit $100 billion in annual recurring revenue, after closing another hosting deal worth about $1.11 billion a month starting December 1st of this year, or roughly $13 billion of additional ARR. Yorkville Ives analyst Dan Ives said the deal provides firepower for SpaceX's AI buildout, adding that the contracted AI backlog and Starlink's cash generation give the company the capacity to service the leverage; Ives has an Outperform rating and a $225 price target on the stock. Earlier this week, Morgan Stanley's Adam Jonas wrote that SpaceX stock was cheap and that now was the time to buy ahead of the upcoming Starship 15 launch, likely scheduled for late October or November.

Marvell Technology raised its revenue guidance at its investor day on Tuesday, forecasting that revenue driven by data centers could grow to as much as $90 billion by its fiscal year 2031. The company also lifted its total addressable market outlook by billions of dollars, and raised its revenue outlook by billions of dollars for both the balance of this year and next year, citing strong AI demand. CEO Greg Murphy, who has led Marvell since 2016, suggested the growth would continue through fiscal year 2031. Marvell is a major chip seller into the data infrastructure space, with business at Google, Amazon and Microsoft, and its stock is up more than 200% this year, though it has not yet returned to its record high. Analysts responded quickly, with roughly eight notes pushing estimates higher on Marvell.

SpaceX is seeking to issue $40 billion in new debt to purchase Nvidia chips for data centers, according to Bloomberg. The financing would be split into $10 billion in bank loans and $30 billion in investment grade debt, with Pimco reportedly looking at the deal and Apollo leading the financing. The company is said to be targeting a 2027 completion date for the raise. The debt would fund additional Nvidia hardware for SpaceX's Colossus 2 data center in Tennessee, which already runs 110,000 Nvidia GB200 chips and 440,000 Grace Blackwell 300s, with another 220,000 Grace Blackwell 300s coming online and a further 220,000 expected in November.

Jefferies started coverage on Ambiq Micro with a Buy rating and a $90 price target, sending the semiconductor firm into the spotlight on Wednesday. Analyst Kevin Garrigan wrote that Ambiq's SPOT architecture runs standard CMOS at a 2-5x power advantage on edge devices, where available power limits AI, and said the company is moving from a niche ultra-low-power MCU supplier toward an edge AI compute platform. Garrigan noted that Ambiq's top three customers, believed to be Google, Garmin and WHOOP, made up 77.5% of sales in the second quarter, but revenue outside those top three grew 143% year-over-year while overall revenue rose 90% year-over-year. He forecast revenue rising from $72.5M in FY25 to $249M in FY28, a 51% CAGR, with non-GAAP gross margin expanding from 45.0% to 48.7%, operating margin improving from -33.2% to 0.7%, and non-GAAP EPS moving from -$2.62 to +$0.53. Ambiq holds about $367M of cash and no debt, which Garrigan said funds the roadmap through breakeven.

Nvidia is set to benefit from a planned $40 billion SpaceX fundraising led by Apollo aimed at securing exclusive access to Nvidia chips. SpaceX is seeking US$40b in new capital to more than double its Nvidia-powered compute footprint for large scale AI and space data projects, effectively treating Nvidia as the default compute layer for those efforts. Separately, OpenAI has chosen Nvidia hardware over Cerebras to run its new Ultrafast AI model, reinforcing Nvidia's role in large model training and its full stack platform anchored by CUDA and TensorRT. The key test for investors is whether these headlines turn into disclosed, multi year capacity commitments that show up in data center segment demand and long term supply agreements, with concrete contract details from SpaceX and further large scale model deployments from OpenAI that explicitly reference new Nvidia platforms such as Blackwell or Rubin. The developments sit inside a wider build out of AI plumbing across data centers and cloud platforms.