The Launch Board

The Launch Board

Every IPO and pre IPO in the AI economy that is worth watching, classified by filing facts rather than hype. A name moves when its paperwork moves.

Data as of November 2, 2026. Educational only, not investment advice.

How to read this board

  • On the pad: A public filing exists, an S-1 or the local equivalent, but the deal has not priced yet.
  • Liftoff: Priced in the last 30 days, or still inside its first trading week.
  • In orbit: Public for less than 12 months and still tracked. Names graduate off the board at 12 months.
  • In the hangar: Still private: credible reporting of IPO preparation, or a plausible candidate by scale alone. Each card says which.

The source labels

  • Filed: Taken from a filing, an exchange record or a company release.
  • Reported: From named financial press or established IPO trackers, not from the company.
  • Rumoured: Market talk with thin sourcing. Carried with a warning, never as fact.

On the pad

A public filing exists, an S-1 or the local equivalent, but the deal has not priced yet.

Anthropic

ANT · Nasdaq Global Select Market
AI labsFiledUnited States

Last private mark: $965 billion post money (Series H, May 2026). Filed terms: none yet. Bankers are reported discussing a $1.5 trillion to $2 trillion valuation.

Why people are watching: It would be the first US listing of a frontier model company, and at reported targets the largest IPO on record, ahead of SpaceX in June 2026.

What could go wrong: The filing shows very large paper losses and hundreds of billions in compute commitments that cannot be cancelled. If public investors price the compute bill instead of the growth, the deal can shrink or slip.

Nscale

NSCL · New York Stock Exchange
Cloud and data centersFiledUnited Kingdom

Last private mark: $14.6 billion (Series C, March 2026). Reported target: $30 billion to $35 billion, with some investors favouring $16 billion to $19 billion.

Why people are watching: The first European neocloud trying to list in New York, and its prospectus puts hard numbers on how much of the AI buildout rests on contracts with two customers.

What could go wrong: The gap between $103.4 billion of paper contracts and 55 megawatts live is the central argument. A Norwegian site runs six months behind schedule and Microsoft exited a planned UK project over power access.

SB Energy

SBE · Nasdaq Global Select Market
Energy and powerFiledUnited States

Reported target: about $50 billion, with a raise of $5 billion to $7 billion discussed. No price range filed.

Why people are watching: The purest public bet yet on the power side of the AI buildout, with OpenAI as tenant, investor and warrant holder at the same time.

What could go wrong: Almost everything being sold is contracted rather than built. If timelines or the OpenAI relationship wobble, the valuation discussion restarts from a much lower base.

Syntiant

SYTN · Nasdaq
Chips and hardwareFiledUnited States

No price range filed. Tracker coverage cites a targeted raise of about $300 million on trailing revenue of about $270 million.

Why people are watching: A rare edge AI chip name at the small end of the board: processors that run models inside earbuds, wearables and drones rather than in data centers.

What could go wrong: A filing that sits unpriced for a full quarter usually means price discovery is hard, and small cap chip deals compete for attention with the mega deals on this same board.

Liftoff

Priced in the last 30 days, or still inside its first trading week.

No liftoff this week. The Liftoff class is empty on purpose: no board company priced in the last 30 days or sits in its first trading week. Anthropic is expected to be the next one, reported for the week of November 16, 2026.

In orbit

Public for less than 12 months and still tracked. Names graduate off the board at 12 months.

SpaceX (including SpaceXAI, the former xAI)

SPCX · Nasdaq
Consumer and otherFiledUnited States

Listed June 12, 2026 at $135 a share, a valuation of about $1.77 trillion. The raise size is contested: $75 billion in tracker tallies, $85.7 billion in one press account.

Why people are watching: The largest IPO on record, and the only public vehicle bundling rockets, satellite internet and a frontier AI lab.

What could go wrong: A listing valuation near $1.77 trillion leaves no room for ordinary disappointment, and the AI division burns cash on compute while the launch business carries its own execution risk.

Cerebras Systems

CBRS · Nasdaq Global Select Market
Chips and hardwareFiledUnited States

Priced May 13, 2026 at $185 a share. Closed at $109.18 on October 30, below the IPO price.

Why people are watching: The largest pure AI chip listing of the year and the clearest public test of whether investors will fund a Nvidia challenger built on a wafer sized chip.

What could go wrong: A large share of contracted obligations traces to one customer relationship, OpenAI, and the stock has already shown how fast scarcity value drains once lockups open.

Zhipu AI (now Z.ai)

2513.HK · Hong Kong Stock Exchange
AI labsFiledChina

Listed in Hong Kong on January 8, 2026 at HK$116.20. Closed at HK$460.80 on October 22, about four times the offer price and about 84 percent below its peak.

Why people are watching: The first large language model company to list anywhere in the world, beating every US lab to the market.

What could go wrong: Entity List status restricts access to advanced chips, and the company has previously limited new signups when compute ran short.

MiniMax Group

0100.HK · Hong Kong Stock Exchange
AI labsFiledChina

Listed in Hong Kong on January 9, 2026 at HK$165. Closed at HK$160.70 on October 22, just below the offer price after peaking at HK$1,330.

Why people are watching: A consumer AI lab that listed a day after Zhipu and briefly traded at a market value above HK$100 billion.

What could go wrong: Reported 2025 revenue of about $79 million against a market value that spent months above 150 times that figure, with thin gross margins for a model company.

Xanadu Quantum Technologies

XNDU · Nasdaq
Chips and hardwareFiledCanada 🍁 Canada

Public since March 27, 2026 via a SPAC merger, not a traditional IPO, dual listed in Toronto and on Nasdaq. The merger valued it at $3 billion pre money.

Why people are watching: The first publicly traded pure play photonic quantum computing company, designing its chips in Toronto.

What could go wrong: Quantum revenue is still small and milestone driven; the company stages fault tolerance as a 2028 to 2029 goal, so the public price rests on technical delivery years away.

Fervo Energy

FRVO · Nasdaq
Energy and powerFiledUnited States

Priced May 12, 2026 at $27 a share. Closed at $13.75 on October 9, about half the IPO price.

Why people are watching: The first large listed pure play on enhanced geothermal, and its headline customer story is now data center power rather than utility power.

What could go wrong: Its plant can be built on time and still be stranded by third party transmission lines, the exact risk now in shareholder litigation.

Standard Nuclear

STDN · New York Stock Exchange
Energy and powerFiledUnited States

Priced July 15, 2026 at $15 a share, below the filed $18 to $21 range, raising $150 million.

Why people are watching: The first pure nuclear fuel IPO of this cycle, reaching the public market about two years after founding.

What could go wrong: No small modular reactor is commercially operating in the United States yet, so its customers timelines are its timeline.

Unitree Robotics

688836 · Shanghai Stock Exchange, STAR Market
Consumer and otherFiledChina

Listed in Shanghai on August 19, 2026 at RMB 150.80. Traded near RMB 426.98 on October 16, about 183 percent above the offer price.

Why people are watching: The first mainland listed humanoid robot company, shipping thousands of robots a year at prices far below Western rivals.

What could go wrong: Its filings show most humanoid revenue still comes from research and education buyers, and profit excluding one time items fell in the first half of 2026 even as revenue grew.

In the hangar

Still private: credible reporting of IPO preparation, or a plausible candidate by scale alone. Each card says which.

OpenAI

AI labsReportedUnited States

Last closed mark: $852 billion post money (March 2026 round). A further round at a reported $1.25 trillion is in talks, not closed.

Why people are watching: At $852 billion it is the second largest private company story in the world, and Anthropic pricing first will set the reference price for its own eventual deal.

What could go wrong: Reported figures show deep operating losses against the valuation, and reporting describes an internal bar of listing only at $1 trillion or more. If public markets will not pay the private mark, the hangar stay gets longer.

Databricks

AI labsReportedUnited States

Last mark: $190 billion (round closed August 13, 2026).

Why people are watching: The largest private enterprise software company in the AI stack, and by its own reporting cash flow positive, which makes it the cleanest eventual listing story on this side of the board.

What could go wrong: Private rounds at rising marks reduce the need to list at all. The longer it waits, the more the eventual IPO price has to justify a $190 billion starting point.

Moonshot AI

AI labsReportedChina

Last reported mark: about $50 billion (final pre IPO round, early October 2026). Reported raise target: about $3 billion.

Why people are watching: It would be the third Chinese lab to list in Hong Kong, and the first to price after investors have seen how Zhipu and MiniMax traded.

What could go wrong: Both listed peers fell more than 80 percent from their peaks within months. Moonshot prices into that record, plus US scrutiny of Chinese model distillation practices, which the company denies.

Cohere

AI labsReportedCanada 🍁 Canada

Last mark: C$35.7 billion, about US$25.5 billion (Series E first close, October 29, 2026).

Why people are watching: Canada's flagship AI company, selling sovereign AI to governments and regulated industries, now with the Canadian state on the shareholder register.

What could go wrong: Reported revenue is small against a C$35.7 billion mark, and the planned Aleph Alpha combination adds cross border integration risk.

Tenstorrent

Chips and hardwareReportedUnited States 🍁 Canada

Last mark: $6.2 billion post money (Series E, October 19, 2026).

Why people are watching: One of the most decorated chip architects alive runs it, and it is the best funded independent attempt at an open alternative to Nvidia hardware.

What could go wrong: Revenue is undisclosed, manufacturing depends on Samsung, and public investors have already marked down Cerebras, its closest listed cousin, by roughly two thirds from its debut close.

Stripe

Consumer and otherReportedUnited States

Last company mark: $159 billion (employee tender, February 2026). Secondary platforms implied roughly $170 billion to $184 billion later in 2026.

Why people are watching: The longest running will they or won't they story in fintech, with each annual tender resetting the price without a listing.

What could go wrong: Nothing forces this listing. Tenders can substitute for an IPO for years, so the risk for watchers is time, not collapse.

Waabi

AI labsReportedCanada 🍁 Canada

Verified fact: a $750 million Series C closed in January 2026. A valuation near $3 billion appears in secondary coverage only.

Why people are watching: A simulation first bet on autonomous driving from one of the field's best known researchers, Raquel Urtasun, with Uber, Volvo and Nvidia's venture arm as backers.

What could go wrong: Autonomous trucking timelines have slipped across the industry for a decade, the robotaxi capital is milestone based rather than in hand, and disclosed road operations remain one Texas corridor.

DeepSeek

AI labsRumouredChina

Reported only, thin sourcing: a first outside round in June 2026 was said to value it above RMB 350 billion, about $49 billion. Carry with care.

Why people are watching: Its open weight models reset global expectations for training cost in 2025, and it remains the largest unlisted Chinese lab by reported value.

What could go wrong: Founder control and a research first culture may keep it private indefinitely, and US export controls shape its compute access.

One honest gap: the Security layer has no entries this cycle. No security company of AI economy scale filed or listed inside the board windows, so the layer stays on the board, empty, rather than being filled with a weak name. Layers follow the Atlas maps; the one deliberate non AI entry, Stripe, says so on its own card.

Left the board

xAI (now SpaceXAI)

Removed from the pre IPO watch: it no longer exists as an independent company. SpaceX acquired xAI in an all stock deal on 2026-02-02 at a reported $250 billion value for xAI, four weeks after xAI closed a $20 billion Series E at a reported $230 billion valuation in January 2026. Renamed SpaceXAI in July 2026. Exposure is now through listed SpaceX, Nasdaq: SPCX, which is on this board in orbit.

CoreWeave

Listed on Nasdaq in March 2025 and has passed the board's 12 month orbit window. Graduated, not dropped.

Data as of November 2, 2026, from SEC filings, exchange calendars and company announcements. Rumoured items are labelled as such. The board classifies facts only: no ratings, no price targets, no buy or sell language. Educational only, not investment advice.