Bernstein and Deutsche Bank both raised their price targets on Space Exploration Technologies on Monday, arguing that Wall Street is still undervaluing the company’s artificial intelligence infrastructure ambitions.
Bernstein lifted its target to $268 from $248 while keeping an Outperform rating, a move that implies roughly 81% upside from the prior close. Deutsche Bank, also rated Buy, took its target to $250 from $230, suggesting nearly 70% appreciation.
The core of the upgrade case is land-based AI computing. Bernstein analyst Douglas Harned said the firm’s revised model is driven chiefly by SpaceX’s plans to build terrestrial AI capacity, which he described as a meaningful addition to the company’s overall revenue trajectory. He added that the argument for a multi-trillion-dollar valuation ultimately hinges on whether orbital data centers succeed.
Harned now projects SpaceX will bring 13 gigawatts of ground-based computing capacity online by the end of 2027, five gigawatts more than his earlier forecast, rising to 20 gigawatts in 2028. He also marked up his assumptions for what the company can charge for compute. SpaceX has said its pricing sits above the $30 to $50 per watt range seen in recent transactions.
The buildout will not come cheap. Bernstein estimates each gigawatt of capacity requires about $35 billion in capital spending. Deutsche Bank’s Edison Yu put SpaceX’s compute cost at roughly $35 billion per gigawatt against an industry standard of $50 billion, a gap that could hand the company a durable cost advantage over rivals.
Yu said estimates for the next several quarters may still be too low and that the market continues to underappreciate SpaceX’s ability to convert a compute edge into better inference and intelligence offerings. He expects the company to add about 10 gigawatts of incremental capacity by the end of next year.
Orbital data centers represent the longer-dated piece of the story. SpaceX intends to launch its first units in late 2027 and start operations in mid-2028, a timeline contingent on Starship achieving reusability so equipment can be lofted more frequently and at lower cost. Harned noted the topic did not come up on the company’s most recent earnings call, though management did provide Starship updates, with Flight 15 expected within weeks and an attempt to catch both the booster and the spacecraft. A successful mission could allow SpaceX to reuse Starship for the first time on Flight 16.
“Reusability is key to making orbital data centers successful,” Harned said.
Management has set an ambitious internal bar, targeting annualized revenue of $500 billion to $1 trillion by the end of 2027. Bernstein’s own model lands well below that at $424 billion for full-year 2028, though Harned said the stock can still be attractive without the company hitting its stated goal.
Sell-side sentiment is broadly constructive. TipRanks data show 29 Buys, five Holds and one Sell over the past three months, with an average target of $250.67. LSEG figures cited by CNBC put buy or strong buy ratings at 33 of 43 covering analysts. The shares have been roughly flat over the past month.
FirmRatingNew TargetPrior TargetImplied UpsideBernsteinOutperform$268$248~81%Deutsche BankBuy$250$230~70%
Note: Implied upside calculated against the prior trading session’s close.
For investors, the debate now turns on execution rather than narrative. The gap between SpaceX’s revenue ambition and Bernstein’s forecast leaves room for further estimate revisions if terrestrial capacity comes online on schedule, while the capital intensity of roughly $35 billion per gigawatt means funding needs will scale alongside growth. Orbital computing remains a binary catalyst tied to Starship’s reusability milestone, and any delay there would shift the valuation burden back onto the ground-based business.