Mega-IPO wave / liquidity Drain ▼

Singh: euphoria is when companies tap markets (the 2021 SPAC/SaaS parallel) — SpaceX ~$75–86B + OpenAI ~$60B + Anthropic ~$60B + Google's $80B secondary ≈ $280B of equity supply by Q3, sapping institutional and retail demand from the rest of the tape just as mutual-fund cash sits under 3% and FINRA scraps the pattern-day-trader rule (suspicious timing). Watch the 6-month unlocks — insider supply hits ~180 days after each debut. Paulo Macro (SPCX-IPO eve): equity funding costs are tightening as leverage demand collides with the $200B of supply — really $800B counting the unlocks — and SPCX may "close the ECM window for good all by itself," stranding Anthropic/OpenAI (whose mooted price war per the WSJ is the tell the exponential story is over). McDonald (Jun 11): the sell-off that started Friday is the indigestion — liquid Mag-7 stock is being sold to make room (record S&P vs equal-weight divergence), convertible issuance is surging (CFOs, "the smartest sellers in the world," dumping equity like Q4-2021), and SpaceX's unlock schedule is "much more aggressive than previous IPOs." Townsend: the real story is ~$3T of restricted shares unlocking by end-2027 (Facebook fell 40–50% in year one as its lockups expired); expect to buy SpaceX ~50% off within a year. Gundlach (Jun 12): the largest IPOs in history (SpaceX touted at ~$1.8T, ~4× oversubscribed, plus OpenAI/Anthropic) cluster near market tops — that mega-cap privates "decided now is a good moment to be selling… is not suggestive that these stocks are cheap"; some of the tape's weakness is funds raising cash to take the SpaceX deal down. App Economy (May 26): the SpaceX S-1 sums-of-the-parts to ~$1.0–1.9T vs the $1.5–2T pitch (~90× 2025 revenue) — Starlink is the 63%-EBITDA floor, AI the $6.4B-loss sink — so "IPO = It's Probably Overpriced"; belongs on the watchlist, but "I'd rather watch this one trade for a few quarters than chase it on day one." Paulo Macro (Jun 5): hundreds of billions of summer equity supply will need funding from grossed-up hedge funds into illiquid markets, and ECM books run "last in, first out" to make room — the buckling of SPCX/CBRS/QNT is the "IPOs forewarn market rollovers" tell. Hay (Jun 4, via Deluard): ~$1.5T of all IPOs since 1792 vs ~$4.6T slated this summer — the real float is <10% but index-inclusion forces operators to buy the new names, pressuring incumbents (only ~5% of the S&P is at new highs). Eisman (Jun 12): tallies ~$360B of AI-era equity supply to absorb (SpaceX ~$75B + OpenAI ~$100B + Anthropic ~$100B + Google's $85B) — and the forced-buying mechanism is the danger: active managers pre-position for the eventual SpaceX/OpenAI/Anthropic index inclusion "regardless of merit," a "required uniformity" that raises the risk everyone underperforms together. Polomny (Jun 13): same supply-drain read — net equity issuance already at highs before SpaceX, with Berkshire's record cash clustering near prior issuance peaks ("usually around tops"); the IPO machine exists to "offload onto retail 401ks" via index inclusion. Singh (Jun 14): the IPO landed — SpaceX listed at ~$2.1T on a 7% float and promptly became a "sell the news" that crushed the listed space/pre-IPO complex (VCX all-time lows, SPCE/PWRL unwound); the supply drain now plays out into the 6-month unlocks, and a $6B SoftBank margin-loan refusal against its OpenAI stock is the first valuation crack in the private-AI names. McDonald (Jun 18): frames SpaceX as the late-cycle credit risk itself — at ~$2T (day-four IPO up 50%, worth more than Amazon) its bankers' $1.3T five-year revenue, if real, would raid Mag-7 cash flow already committed to $5T of on-Earth data centers; with $800B of off-balance-sheet data-center financing, "SpaceX could create a credit crisis" — the AOL/RJR-Nabisco hubris moment, triggered when the funding window closes. Paulo Macro (Jun 17): SPCX is "sucking the oxygen out of the equity room" — options listing + Russell inclusion (days away) + Nasdaq (early July) + the "narrowing in" of a final top set up "something truly acrobatic" (a Volkswagen-2008 squeeze ×10-20 that "doesn't end until SpaceX is the world's largest company"); it's also coming to the IG bond market with $20-25bn of debt, ~25bps tighter than ORCL — a credit-froth tell. Eisman (Jun 18): SpaceX IPO'd +19% day one ("not particularly impressive"), then +20% Monday / +5% Tuesday to a ~$2.5T cap at >100× trailing revenue; Anthropic & OpenAI IPOs are next in the fall. App Economy (Jun 19, "How to Invest in IPOs"): 40 years of Jay Ritter data say wait — from the first-day close, 50% of operating-company IPOs (≥$100M sales) are negative over five years and only 24% become multi-baggers, with underperformance worst 6–24 months in (lock-up expiry); the 2026 wave (SpaceX +50% in three days, Anthropic/OpenAI filing) is a market-temperature readout, not a buy signal — "watch list, not impulse buy." Polomny (Jun 27): SpaceX is 'not investing' — lockups and IPO mechanics make retail the 'exit liquidity' for the VCs/insiders who bought in cheap, and forced index-inclusion buying offloads it into 401(k)s; Starlink is 'a step behind 5G' and the Mars / data-centers-in-space pitch is S-1 hype. Retail 'piling into semiconductor ETFs' at record inflows are 'the bag holders.' Halftime (Jul 15): the wave builds — Anthropic is lining up investor meetings ahead of a possible ~$1T listing "as soon as October" (the first pure-play AI lab to go; OpenAI has also filed confidentially), while SpaceX closes below its $135 set-price IPO for the first time: Terranova — the appreciation was "built upon scarcity" and the debt market was "the adult in the room"; July's thinning liquidity plus lockup/SPV supply is repricing it. Pabrai (jun-30, rec. jun-8): SpaceX's coming mega-IPO is "a very big shiny object" — the next destination for money rotating out of AI (as Bitcoin money rotated into AI before it); his stance is to stand aside, not participate: "IPOs are like lottery tickets — many will play, few will win." Smead (jun-10, back-fill): Newton's third law of liquidity — a mega-IPO/secondary wave (SpaceX) forces passive indexes to SELL their largest constituents (GOOGL/META/MSFT/AMZN/NVDA, the forced-selling epicenter) exactly as those names cut buybacks and turn equity raisers (Google raised; Meta will): the dilution moment of the cycle. Tops = running out of buyers, and these IPOs "meet the exhaustion"; the run-out-of-sellers season that follows is the bear market. Precedent: internet capex halved 2000–02 even as usage exploded — technology proliferates because price falls. Finucane (Jul 16/17): skips the mega private IPOs at 20–70× revenue (vs Google's 8.5× in 2004) — SpaceX priced ~58× next-year's revenue and is already down ~16% from its first-day close, a bullish sell-side report concedes no SpaceX free cash flow until at least 2035, and hot IPOs typically halve at some point in year one. The patience rule: if you really want one, wait the first year (Oxbow watched Airbnb 4–5 years and bought ~$120 vs its $145 IPO). Salzman (Jul 16, Barron's): Standard Nuclear (STDN) downsized its IPO and broke its $15 price (opened $13.50, closed $12.26) as investors soured on new-nuclear stocks — a cooling-appetite datapoint at the speculative edge of the IPO wave. Singh (SSR, Jul 19): the market is preparing for the SpaceX lock-up unlock (August) — his trade is short SPCX against long SATS/EchoStar, with the spread compressing as SpaceX fell more. Codex (Jul 31): Westinghouse has confidentially submitted a draft Form S-1 for a proposed IPO (no share count or range set) — it fits Brookfield's pattern of monetizing a built-up asset and rhymes with the US-government partnership terms, where the state's participation interest and forced-listing option vest at a $30B+ valuation before 2029. Mart declines to invent a valuation and would rather Westinghouse stay on Cameco's books — "one nuclear power and fuel cycle giant" commanding a higher multiple for CCJ. Salzman (Jul 31, Barron's): a potential blockbuster nuclear listing joins the wave — while the theme's prior nuclear deal (Standard Nuclear, Jul 16) downsized and broke price, so supply is accelerating into a bid already showing strain. 2026-AUG-03 (Joseph Carlson): the SpaceX unlock is a standing overhang for the rest of 2026 — relaying Tom Lee, "SpaceX will continue to have that unlock throughout the rest of the year. That'll put downward pressure on the stock market this year as SpaceX continues to sell off more and more shares." Clearing it is one of the two clouds Lee expects lifted by 2027 (the other being the market testing of the new Fed). 2026-AUG-02 — Mohnish Pabrai (New Money): on the record-size SpaceX IPO — the business is "phenomenal" and its doing well is "almost a no-brainer" ("do not short a superhuman"), but as an investment it goes in the too-hard pile: asteroid-mining offering documents, unknowable economics. "I don't need to make that bet." 2026-AUG-10 — Le Shrub (Paulo Macro, Fly on the Wall): the mega-IPO is now a passive-flow capture device — mark up privately, list into the index "at a stupid valuation and the passive flows will maintain that stupid valuation"; SpaceX the exemplar of "trillion dollar companies added to the index on IPOs." Singh (Aug 9): SpaceX is down ~40-50% from its IPO and "has dragged down a lot of the space stocks with it" (ASTS, RKLB, SATS). Q2 missed — revenue $7.8B (+92%), EPS −$0.09 — despite Starlink subscribers doubling to 12M and a $47B backlog, with Musk targeting 2 GW of compute by year end scaling toward 10 GW, all debt-funded. SSR closed its SATS-long/SPCX-short pair and stays sidelined ("a lot of the float is still unlocking, so we haven't added back SATS yet"); the mark-down also forces Alphabet to reverse part of its Q2 paper gain. 2026-AUG-05/06/11 — CNBC Halftime, the engineered-IPO cycle: SpaceX's inaugural report (revenue +92% to $7.8B, AI revenue +247%, $18.4B capex) sends it −9% into the first lock-up tranche, ~50% below the highs; Simpson holds a half-percent position taken at 160 and deliberately has not added: "just because a stock pulls back doesn't mean you need to add… I want to see how it trades with these new shares as part of the marketplace." Aug 6 (Brennan/Brown): the lock-up opens — available shares +140% to over 1.5bn (50% free float by January per Bernstein), offset by index-methodology changes that make a bigger float mean a bigger weight and forced index buying at the Nasdaq-100's early-September adjustment; Brown: the IPO "was engineered to perfection — immediate index inclusion, a 5% float, insiders locked up. This is the first click anybody's going to get," and $86B raised with a fifth spent in one quarter means another raise is coming. Aug 11 (Rooney/Sechan): Anthropic's IPO could come as soon as October — bankers engaged, WSJ reporting investor questions on a revenue run rate last topped at $47B amid cheaper Chinese open-source models; OpenAI's Brad Lightcap departs as power consolidates under Altman/Brockman ahead of a listing. Sechan on the pattern: "you get a big buy-up and then they re-rate. Meta did that, SpaceX did that… large engineered IPOs with lockups that are unique, and markets take time to find price discovery." 2026-AUG-07 — Salzman (Barron's): the index-access lever named explicitly — around SpaceX's dual listing on Nasdaq's New York and Texas venues, Nasdaq created a "fast track" into the Nasdaq-100 after just 15 trading days, which AFR's Natalia Renta ties to venue competition ("and then all of a sudden, their exchange gets the listing") and to average investors taking "inordinate risks." Nasdaq says the program makes the index better reflect its market and that the decision predated SpaceX. AUG-16 (Jay Singh, SSR): the pipeline is bifurcating. On one side, private AI is preparing to list — combined OpenAI + Anthropic annualized revenue went from $29B to ~$100B with Anthropic profitable in Q2 and, per friends of his, "looking to actually IPO in October" — and Unitree's $9B Shanghai IPO is oversubscribed ~8,000% by retail. On the other, the sponsor-backed channel is shut: "since 2022, only 70 private equity backed companies have actually gone public over the last four years… from 2017 to '21, there were 424," with less than 10% of PE exits now IPOs. 2026-AUG-19 — Halftime (Kate Rooney): The two frontier labs are now both on file with the SEC, and the first side-by-side is unflattering to OpenAI: per the WSJ its operating loss grew to $12.3B in Q2 from ~$9B in Q1, with losses outpacing revenue growth (revenue just under $7B, +18% q/q). Anthropic "just has the better margin profile… was profitable at least on an EBITDA basis" on $11.5B in the quarter; annualised, OpenAI ~$40B vs Anthropic ~$65B. Two discounts stated on air: the leak is backward-looking to end-June (July revenue +20%, enterprise +32%, confirmed by Greg Brockman), and "we don't know if these are apples to apples… the accounting can differ." The comparison gets settled in the S-1s. Kedrosky (Meb Faber #648, Aug 28) sizes it at $4T rising to ~$5.5T from SpaceX, Anthropic and a couple of peers alone — more than all post-World-War-II US IPOs combined, and the one bubble ingredient absent until now. The mechanism matters more than the number: large funds hold no spare cash (cash is a performance drag), so allocations are funded by selling, and what gets sold is predictable — the most liquid, the most overlapping with what's being bought, and perversely the best-performing holdings (locking gains for the quarterly letter). Because nobody raises billions the night before, the selling starts weeks-to-months early and quietly, which is why he modelled it in March/April and expected pressure on the liquid winners in April–June. He believes that anticipated funding flow "became a pin" under the Situational Awareness implosion — the fund was long exactly the names destined to be the funding source. Corroborating regime shift in listed names: Oracle moving from buybacks toward share issuance and dilution. Chris Mayer (Value After Hours, 2026-AUG-25) confirms the structural side of it as a durable change rather than a cycle artefact: companies staying private for longer is "a real phenomenon — IPOs are much more mature now and bigger than they were," so the public market increasingly buys businesses at a later, larger, more fully-priced stage than in prior cycles. Steve Eisman (The David Lin Report, 2026-AUG-28) treats the coming lab IPOs as a disclosure event rather than a supply event — the moment a risk that is currently unobservable becomes measurable. "I'm hoping to get some clarity on Anthropic when they put out their S-1. Whether they'll give it or not, I don't know." Prediction markets price Anthropic at 93% to list before OpenAI, so Anthropic's filing is likely the first real look at the numbers the whole AI chain depends on. (New Money clip, 2026-AUG-15): "a key thing to monitor to determine a catalyst for a real sustained sell-off is the health of Anthropic and OpenAI" — both private today, which is precisely why the filings matter. 2026-AUG-31 — Jeff Keller (Other People's Money): Anthropic's S-1 lands within a week or two. He expects little new information (the numbers are broadly leaked) but sees public labs as a net positive — replacing third-hand ARR leaks with real metrics and damping narrative volatility. On a SpaceX-style capital vacuum (space also-rans were sold the moment SpaceX became buyable): possible but not worrying — equity issuance into real businesses at realistic prices "is less capital sucking than we need to put $10 billion into Nikola." Semi-trade positioning has already cleaned out; the market "will want lab exposure" and Anthropic "is going to trade at a crazy price… a lot of enthusiasm for it." Slegers 2026-JUL-12: names the pipeline absorbing the market's attention — SpaceX, OpenAI, Anthropic, Stripe — and separates it from the memory trade as "a different kind of problem": these "lose money every single month," where the chip makers are over-earning. His framing of the market as "euphoric about unprofitable IPOs and cyclical commodity stocks" is the clearest statement in the archive that the two speculations are distinct trades that will not unwind together. Slegers 2026-JUN-18: the SpaceX listing priced and dissected — IPO $135, debut $150, $191 within a week for a $2.5trn market value against a $1.75trn IPO valuation, i.e. 90x revenue ("higher than Palantir at 75x, Nvidia at 20x, Tesla at 16x") on 2025 revenue of $18.7bn and a $4.9bn loss, with xAI burning $7.7bn in one quarter. To justify 20x sales needs $88bn of revenue — "still 6 years away" at 30% growth; Morningstar's fair value of $780bn is less than half the market price; underwriters earned "over $500 million just in fees." The rule drawn: "IPO… It's Probably Overpriced." App Economy Insights (2026-SEP-11): Anthropic's pre-IPO optics turned. Researcher Jacob Coxon quit warning labs are 'gambling with our lives,' and alignment lead Evan Hubinger put the odds of AI killing all humans within a decade above 10%: 'not exactly ideal PR ahead of a potential IPO.' CNBC Halftime (2026-sep-11): Harrington names the next unpriced disclosure event for the whole AI complex — "I'll be curious once we see the S-1s from OpenAI and Anthropic... what if there's a little bit of pullback? I think the trickle from a marginal pullback would be pretty painful." First audited financials for the theme's demand side, arriving into a tape where one customer (OpenAI) already sits inside another company's backlog valuation (Oracle's ~$700B). Jay Singh (SEP-13): expects Anthropic's IPO "till October at the earliest, next year is the most likely," against investors pushing an October 26th listing near $2T; "OpenAI definitely cannot do an IPO this year"; doubts an Anthropic mark above $1.7T soon. David Woo (2026-SEP-14): wouldn't buy the Anthropic IPO "not at two trillion dollars" — its secondary shares traded nearer $1.3T, so insiders who could wait six weeks are selling (nobody sold six weeks before SpaceX's IPO); the listing is "a referendum of the entire AI trade" — if it goes poorly "the whole AI trade could basically collapse." 2026-SEP-14 (CNBC): the bear case flipped from too much IPO supply to deals not happening — Goldman trades down as the proxy; Altman says an OpenAI IPO now would be "ill advised" (2027); Rooney's sources say Anthropic has had back-to-back quarters of adjusted profitability; Santoli and Weiss say the S-1s (profitability, agent product-liability risk factors) are the test. 2026-SEP-15 (App Economy Insights): SB Energy files its S-1 at a reported >$50B valuation on $139M of H1 revenue — a Nasdaq controlled company (SoftBank >50% of votes), the IPO 'part of the financing model itself'. Verdict: 'It's Probably Overpriced'; wait for operating campuses. Porter Collins (Eisman Ep 75, 2026-SEP-14): Anthropic and SpaceX coming public are "new stock of supply… that's what usually has killed markets… in 2000, in 1929," against passive 401(k) demand that only buys on inflows. Sep-10 (Paulo Macro / Le Shrub): the Anthropic IPO “has to succeed,” so expect a month of bullish AI stunts; Le Shrub says it gets done, possibly “the absolute meme top.” Paulo doubts the window (DeepSeek at 1.4% of the cost, data-center opposition, midterms) and notes a pulled deal would itself mark “a near-term low in the market.” Joseph Carlson (Aug-27, 2026): SpaceX at ~$1.84T with two public quarters is unmodellable — AI compute ate $15.8B of $18.4B quarterly capex — so only a small "buy into Elon's vision" bet. Andrei Jikh (2026-SEP-16) flags AI-doom messaging as pre-IPO marketing. The viral Anthropic-resignation thread (100M+ views) came as Anthropic is "about to" go public, and "it's really hard to justify an AI chatbot at a trillion dollar valuation, but the most dangerous technology in human history… makes it a lot more justifiable." Michael Green (2026-SEP-11): the SpaceX listing was index arbitrage, not price discovery. Fast-track Nasdaq-100 inclusion plus roughly 3:1 float magnification made the passive bid the insiders' exit while retail was locked up for 30+ days. It roughly quintupled into the rumored inclusion (the Tesla/S&P 2020 analog), passed $3T, then fell to about $1.25T as net selling began. It repeats the 2020-21 SPAC trick: CRSP could be forced to buy in 5 days while insiders were locked for 20, until CRSP closed it in Sept 2022. Now it is back via fast-track IPOs, which only mega-caps can use. Sep 17 (Salzman, Barron's): the nuclear leg of the IPO wave stalls — Holtec withdrew a ~$900M / $10B listing on "market conditions" after STDN broke issue in July, and dual-class terms (CEO majority control via Class B shares) "gave some shareholders pause"; the withdrawn supply lifted listed peers XE and OKLO +12%. Deluard (2026-SEP-17, Risk Takers): the SpaceX IPO (June 2) coincided with the Nasdaq's high that day (30,666 vs 28,845 now) on a "$30 trillion TAM" pitch; Anthropic pushing its IPO back (S-1 still undisclosed, profitable only "if you remove the cost of training") and OpenAI saying it may not list at all read as late-cycle tells. Halftime (2026-SEP-17), Santoli: after SpaceX the worry flipped from equity oversupply to "what if we don't get OpenAI and Anthropic IPO" — "they need the money to pay for the stuff they ordered from all the hardware makers." Ethridge: an AI-lab CEO voicing safety concerns about "the product that we were planning to IPO in a couple of weeks" doesn't help; Goldman is his final trade as the likely lead on the Anthropic prospectus. 2026-SEP-18 (Eisman): Anthropic "is going to go public this year… maybe in a month or two" - which is why it "can't slow down"; "it's unclear to me how to keep a shell game going in an IPO process that requires transparency." George Noble on The Real Eisman Playbook Ep 76, 2026-SEP-21: SpaceX's float has unlocked from ~5% to 20-25% of shares, with "every few weeks 7% more"; its extreme valuation rests on the narrative plus "the incredibly small float, which is in the process of changing," and early holders and endowments whose positions got too big "have to sell." Eisman: Anthropic needs its S-1 out "fast because token maxing is ending."