AI Markets: OpenAI at $1.25 trillion, SpaceX and cell towers, neocloud credit

OpenAI trims its valuation ahead of closing its funding round, and SpaceX is reportedly eyeing 80,000 cell sites. Meanwhile, neoclouds are piling on debt and Tesla is firming up its robot timeline.

OpenAI lines up $30 billion at a $1.25 trillion valuation, below its earlier target

OpenAI is reportedly close to wrapping up a raise of up to $30 billion. The valuation being floated, $1.25 trillion, is lower than the $1.4 trillion pitched to investors in early October.

Key points:

  • According to CNBC, the round would reach up to $30 billion at a $1.25 trillion pre-money valuation, or roughly $1.28 trillion post-money if the round is fully subscribed.
  • Investors from the United Arab Emirates and Coatue are said to be leading the round; OpenAI is reportedly aiming to have signatures by the end of next week.
  • Nothing has been announced: every figure comes from people familiar with the matter, and the final amount has yet to be set.
  • OpenAI isn’t publicly traded: there’s no direct access, whether through a PEA or a regular brokerage account.

What has changed since our last edition

Until now, the press had been talking about a round of at least $30 billion at around $1.4 trillion. The new range is therefore about $150 billion lower. None of the outlets we checked gives an official explanation for the pullback. The timeline, on the other hand, is coming into focus: the close is said to be near, while an IPO is still being floated for 2027.

SiliconANGLE describes the round as “bridge” financing meant to carry the company through to its IPO. That’s the press’s interpretation, not OpenAI’s wording.

The link to the stock market

A lower valuation changes the comparison with Anthropic, whose IPO is reportedly being floated for November at around $2 trillion, according to the press. Both labs are raising money at the same time, and investors have to pick.

For a European investor, exposure to OpenAI currently runs through its listed partners, such as Microsoft or SoftBank. Those stocks are exposed to OpenAI alongside other businesses. Their share prices don’t reflect the value of this lab alone.

Sources:

SpaceX reportedly wants 80,000 cell sites to become a fourth carrier

SpaceX is reportedly negotiating access to at least 80,000 cell tower sites in the United States for its Starlink Mobile service. The report comes from Bloomberg and hasn’t been confirmed by any of the parties.

Key points:

  • According to Bloomberg, SpaceX is in talks with tower owners, including Crown Castle, to install its equipment on at least 80,000 sites.
  • That would be roughly half of the 158,500 purpose-built towers in the United States, according to Wireless Infrastructure Association figures cited by the press.
  • Crown Castle jumped about 9% after the close, according to Benzinga.
  • No signed contract is mentioned; the sums discussed are “in the billions.”

Why this is new

Until now, we had covered the space side: the FCC approval for 15,000 satellites and the purchase of 800 MHz spectrum from Grain, signed on October 8 and expected to close in the first half of 2027, pending approval. Cell towers add the ground side. SpaceX would no longer be betting only on satellite-to-phone, but also on a terrestrial network.

Elon Musk said current carriers “don’t understand” what Starlink is preparing, according to Yahoo Finance. That’s an opinion, not a fact.

Potential winners and losers

Tower companies would be on the right side of this: they already lease their sites to multiple carriers. Bank of America believes a real terrestrial network would make using them “almost inevitable.” Carriers AT&T, T-Mobile and Verizon fell on the news, according to PhoneArena.

For European investors, these stocks are all American. They aren’t eligible for the PEA. Keep in mind, too, that SpaceX hasn’t confirmed a nationwide rollout.

Sources:

Neoclouds: Nebius borrows $3.5 billion while Goldman stands by CoreWeave

Debt-funded AI compute providers remain under pressure. Nebius announced a $3.5 billion short-term loan, and Goldman Sachs upgraded CoreWeave to buy despite the stock’s slide.

Key points:

  • Nebius: a $3.5 billion secured facility from Citi, maturing in 364 days unless refinanced.
  • Nebius stock lost 7.23% to $159.17 on October 22, the day of the announcement (TIKR).
  • Goldman Sachs upgraded CoreWeave from “neutral” to “buy” on October 19, while cutting its price target from $139 to $106.
  • According to Goldman, investors are overestimating CoreWeave’s future cost of capital.

A loan to buy time

The Nebius loan funds data centers, chips and general corporate purposes. A 364-day maturity is short for this business. TheStreet notes that the terms suggest the company is “buying time,” while GPU rental prices keep running ahead of plan.

CEO Arkady Volozh said in September that chip supply and demand would eventually balance out, within one to two years. He doesn’t rule out higher prices: “Maybe they’ll go up, who knows.”

Two opposing views

The cautious view: CoreWeave had about $35.6 billion in debt against $5.5 billion in cash as of June 30, and the yield on its 2032 bonds hit 15% last week. Nebius, for its part, had $4.5 billion in debt, less than its cash at the end of September.

The confident view: Goldman analyst Gabriela Borges sees the drop (about 43% since the August 11 earnings) as an attractive entry point. These are analyst opinions, not certainties. CoreWeave and Nebius are American stocks, not eligible for the PEA.

Sources:

Tesla: Musk clarifies the Optimus timeline after mixed results

Tesla beat earnings-per-share expectations, but the stock slipped 3%. The Optimus timeline, meanwhile, is getting clearer.

Key points:

  • Adjusted earnings of $0.48 per share versus $0.45 expected; revenue flat at $28.0 billion (TIKR).
  • Net income down 43% to $787 million, as operating expenses rose 39%, mostly in R&D.
  • 2026 capex cut to about $20 billion, versus more than $25 billion planned, according to the CFO as quoted by TIKR.
  • Stock at $375 at the close, or about 179 times expected earnings (five-year average: 108).

What has changed on Optimus

Our last edition flagged conflicting dates for external sales. Musk clarified that producing 20,000 robots a week would take “probably at least a year, maybe a little longer.” He expects to sort out the remaining design and AI issues around the end of next year. Tesla would then start delivering Optimus to other companies.

The robot is in early production in Fremont. These targets are still executive statements, not results.

The link to the stock market

The market is paying for Tesla’s future, not the current year. At 179 times earnings, the valuation assumes robots and autonomy will deliver on their promises. A slipping timeline therefore weighs more heavily than it would on a company valued on its current results.

The possible tie-up with SpaceX, with special committees already formed, adds another unknown. Tesla is an American stock, not eligible for the PEA.

Sources:

Edition prepared by our monitoring agent (data and sources as of Friday, October 23, 2026), reviewed in line with our editorial policy. General information, not investment advice: investing carries a risk of capital loss.