After SpaceX: the 2026 IPO pipeline, ranked by what's actually filed

$SPCX priced at $135 on June 11, started trading June 12, and closed its first session at $161, up 19%. Roughly $75B raised at a ~$1.75 trillion valuation: the largest IPO in history, about 2.5 times Saudi Aramco's 2019 record. The search query that follows a day like this is predictable: "what's the next big IPO."

The standard answer is a list of ten famous private names. The standard answer is mostly noise. Rumor lists are long; filed S-1s are short. What separates a 2026 listing from a 2027-or-never listing is paperwork, and as of today the paperwork sorts the field cleanly into four tiers.

The wave so far: two debuts, two blowouts

2026's IPO season opened before SpaceX did.

• Cerebras ($CBRS), May 14. The wafer-scale AI compute name raised $5.5B and opened 89% above its $115-125 range. It was the year's largest tech IPO for all of four weeks. The full structural read is in the Cerebras IPO piece, and the live data sits on /stocks/cbrs.
• SpaceX ($SPCX), June 12. Priced $135, closed $161. The offer raised ~$75B on 556.6M shares. One structural detail that resets a lot of stale watchlists: xAI was merged into SpaceX in an all-stock deal in February 2026, so SPCX is also the public xAI exposure. There is no separate xAI IPO coming. What you actually own at this valuation (launch, Starlink, Starship, xAI) is broken down in the SPCX explainer, and the universe page is /stocks/spcx.

Two data points are not a trend, but they rhyme: both deals were heavily oversubscribed, both priced into demand that the offer size did not satisfy, and both paid first-day buyers. That backdrop is exactly why the rest of the pipeline is accelerating its paperwork.

The pipeline, ranked by evidence

| Rank | Company | Filing status | Window | Valuation talk | |---|---|---|---|---| | 1 | Anthropic | Confidential S-1 submitted (June 2026) | As early as October 2026 | ~$900B private | | 2 | OpenAI | No S-1 | Late 2026 / 2027 | ~$852B private | | 3 | Kraken | Confidential S-1 (Q1 2026) | On hold | ~$20B | | 4 | Revolut | Confidential S-1, Nasdaq | May wait 1-2 years | ~$75B | | 5 | Stripe | None ("IPO-ready") | Not committed | n/a (profitable) | | 6 | Databricks | None | Candidate | n/a | | 7 | Anduril | None (raising private) | Further out | ~$60B private |

The tier boundaries matter more than the ranks. Tier one is "filed with a window" and contains exactly one company. Tier two is "filed but parked." Tier three is "could file whenever they want and conspicuously have not."

Anthropic: the only name with a filed S-1 and a date

Anthropic confirmed a confidential S-1 submission in June 2026. Reporting around the filing puts the IPO window as early as October, alongside talks to raise $50B privately at a roughly $900B valuation. That private mark now prices Anthropic above OpenAI, a reversal that happened quietly over the spring.

What makes this listing structurally interesting is not the size. It is that Anthropic would be the first frontier-lab pure play on the public tape. Today, public-market AI exposure is hardware ($NVDA, $CBRS, the AI Compute Accelerators bubble), infrastructure (power, cooling, networking), or diversified mega-caps where the lab is a rounding line ($GOOGL, $MSFT via its OpenAI stake). A frontier lab with its own P&L on a ticker is a new asset class, and index funds, AI-thematic ETFs, and every "picks and shovels" portfolio will have to decide what to do with it.

Open questions the S-1 will eventually answer: revenue scale and mix (enterprise API vs consumer), the cost curve on training runs, and how the Amazon and Google strategic stakes translate into public-market governance.

OpenAI: bigger event, earlier stage

OpenAI is the larger franchise by revenue: roughly $25B annualized against a ~$852B private valuation. But there is no filed S-1, and CFO Sarah Friar has publicly framed late 2026 or 2027 as the realistic window. The corporate-structure conversion that preceded any filing was the slow part; the listing itself is now a sequencing question.

One wrinkle distinguishes OpenAI's path from every prior tech IPO: the US government. At a June 3 White House roundtable, the administration said it would look into taking "a good solid stake" in major AI companies, and reporting since has put OpenAI in active talks about a structure where equity seeds a public wealth fund. The administration has already executed roughly ten government-equity positions in 2026, with the $INTC CHIPS-grant conversion, the DoD preferred stake in $MP, and DOE warrants in $LAC as the live precedents. A sovereign stake negotiated before a listing changes float math, governance, and the narrative in ways no S-1 template covers. The same overhang applies to Anthropic, which was named in the same policy conversation.

The frozen and the patient

• Kraken filed confidentially in Q1 2026 at a ~$20B target and then watched its window close: crypto prices pulled back, volumes softened, and recent crypto-adjacent listings underperformed. The filing exists; the conditions do not. On hold.
• Revolut has a confidential S-1 and a Nasdaq designation at a ~$75B valuation, and reporting suggests it may deliberately sit for another year or two. Filed-and-patient is a real strategy when secondaries keep employees liquid.
• Stripe is the permanent next-year name: profitable, "IPO-ready" by its own framing, with employee liquidity handled through secondary sales. No filing. A company that does not need primary capital lists on its own clock.
• Databricks has no filing and is best treated as a candidate rather than a pipeline entry, perennial 2026 chatter notwithstanding.
• Anduril is still pricing private rounds (~$60B talk) and sits furthest out. Defense-tech demand is real; so is the convenience of staying private while procurement scales.

The mechanics the SPCX debut just taught

Three observable lessons from the last month that apply to every name above:

1. Index inclusion is not automatic. S&P Dow Jones declined a fast-track for SPCX; S&P 500 entry is gated on GAAP profitability and float, realistically 2027, while Nasdaq-100 and Russell entry move faster. Anthropic and OpenAI would face the same gates with their own loss profiles.
2. Allocation goes to institutions; retail gets the open. CBRS opening +89% and SPCX +19% is the cost of buying at the bell instead of the offer. The bigger the brand, the wider that gap tends to be.
3. Lockups are the second event. The SPCX lockup expiry lands roughly December 2026, joining the calendar tracked in the AI IPO lockup piece. Supply that arrives 180 days after a hot debut has repriced more than one of these stories.

What trades the theme today

Until the next S-1 goes public, the pipeline is mostly unownable. What is ownable: $SPCX and $CBRS are on the tape now and mapped in the universe (/stocks/spcx, /stocks/cbrs), the Space / Sat Comms bubble carries the cohort that traded the space theme for years before SpaceX listed, and the AI Compute Accelerators bubble is where a frontier-lab listing would send its capex first.

What we are watching from here, in order: the Anthropic S-1 flipping from confidential to public (that is when a price range and a date become real), any OpenAI filing, Kraken's window reopening with crypto volumes, and the December SPCX lockup. Each of those is an observable event, not a prediction. The list above will be wrong in one direction or another by Q4; the filings will say how.