The Great Equity Supply Test: Can $1.4T in Buybacks Absorb SpaceX Lockups and a $2T Anthropic IPO?

For the better part of two decades, the U.S. equity market enjoyed a quiet structural tailwind: the supply of publicly traded shares was stable or shrinking. Companies bought back more stock than they issued, private equity took companies private faster than the IPO market replaced them, and low interest rates made the whole cycle self-reinforcing. That tailwind is not gone — but it is bending. The question for the back half of 2026 is not whether buybacks still exceed issuance (they do, by roughly two to one), but whether the gap is narrowing fast enough to matter.

The Buyback Machine Still Runs — But the Other Side Is Growing

Goldman Sachs estimates U.S. companies will repurchase approximately $1.4 trillion of shares during 2026, a figure that would offset around $700 billion of primary equity issuance plus additional supply from post-IPO lockup expirations^[1]. S&P 500 buyback growth was running at roughly 11% year over year in the second quarter, and new share-repurchase authorizations have reached a record of nearly $1 trillion year-to-date^[1].

On the surface, that is an overwhelming demand signal. Buybacks at twice the level of primary issuance mean corporate America remains a significant net buyer of U.S. equities. Goldman’s strategists, led by Ben Snider, put it plainly: “Follow-on equity issuance is increasing but represents a return to normal rather than a boom.”^[1]

Yet the composition of that issuance is where the trajectory gets interesting. U.S. companies raised $105 billion through follow-on equity offerings in the year through July — the strongest pace at this stage of a calendar year since 2021^[1]. Total equity issuance, including IPOs, follow-ons, convertibles, and SPACs, reached a record $252 billion in the second quarter, surpassing the prior quarterly high of $234 billion set in Q1 2021^[1].

AI Capex Is Driving the Supply Side

The single most important driver of new equity issuance is artificial intelligence infrastructure spending. AI-related transactions have accounted for approximately 40% of U.S. follow-on equity issuance this year, and Goldman expects that share to increase^[1].

The math is straightforward. Consensus forecasts put hyperscaler capital expenditure at $1.1 trillion in 2027, which would exceed operating cash flow by roughly $150 billion before these companies are expected to turn free-cash-flow positive again in 2028^[1]. If capex comes in above consensus — as many investors expect — the financing gap widens. Goldman’s credit strategists estimate hyperscalers could fund about 35% of 2027 capex through debt, translating into roughly $400 billion of global debt issuance, but equity capital “should also continue to play a role”^[1].

The pattern is already visible. Alphabet has sold over $85 billion in bonds and $20 billion in stock to fund its AI build-out^[2]. Meta and Amazon have each tapped bond markets in recent months, and while they have not yet come to market with secondary stock offerings, Interactive Brokers’ chief strategist Steve Sosnick notes they are increasingly likely to use buybacks merely to offset employee-stock dilution rather than meaningfully reduce share counts^[2].

That is a subtle but important shift. When the largest companies in the market transition from net buyers of their own stock to neutral participants — buying just enough to offset dilution — the buyback tailwind that has supported equity prices for years weakens at the margin. It does not reverse, but the slope changes.

The SpaceX Lockup Test: What Actually Happened

SpaceX’s June 2026 IPO at $135 per share was the largest public offering on record at the time. By early August, the stock had slumped approximately 38–49% from its post-IPO intraday high^[3][4]. The first lockup expiry on August 6 was, in the words of R.F. Lafferty CEO Robert Hackel, “the most talked-about lockup in the history of IPO lockups”^[3].

As many as 912 million of SpaceX’s roughly 13.6 billion outstanding shares became eligible for sale — enough to more than double the public float, and potentially triple it if a price-based early-release provision triggered^[3]. A staggered schedule will free an additional 12.9 billion shares by mid-2027^[3].

The anticipated deluge, however, did not arrive. SPCX shares actually rallied after the lockup expired^[2]. Early investors and employees, many sitting on gains from pre-IPO stakes acquired at a fraction of the IPO price, largely held rather than sold. Some wealth managers reported that insiders they contacted expressed long-term conviction and no urgency to exit^[3].

This is a meaningful data point, but it is one observation, not a pattern. The staggered lockup schedule means additional tranches will unlock through mid-2027, and the selling pressure could simply be deferred rather than absent. Elon Musk’s roughly 42% stake remains locked for a full year under a separate agreement^[3]. Executive officers are subject to longer lockups that generally do not begin expiring until after fourth-quarter results^[3].

The base case here is probably 60/40: 60% that early holders continue to dribble rather than dump, supporting the stock’s stability through subsequent unlock dates; 40% that any rally back toward the $135 IPO price triggers pent-up selling that re-accelerates downward pressure. The key variable is not the number of shares unlocked — it is who among the early backers (Founders Fund, Craft Ventures, Alphabet, Valor Equity) decides to trim, and whether that signals anything about their confidence in the company’s trajectory^[3].

The Anthropic Pipeline: A New Scale of Supply

While SpaceX’s lockup overhang is the near-term variable, the forward pipeline is dominated by the expected Anthropic IPO. Anthropic confidentially filed its prospectus with the SEC in June 2026^[4]. CFO Krishna Rao is now leading early investor meetings that remain high-level — covering Claude models, Claude Code, enterprise positioning, and management — without yet discussing specific financials or a valuation^[4].

According to the Financial Times, investors are floating a $2 trillion valuation, based on their own analyses rather than guidance from the company^[4][5]. That figure rests on Anthropic’s revenue run rate, which crossed $47 billion in May 2026, up from roughly $10 billion in all of 2025^[4]. Some investors expect the run rate to reach $100–120 billion by year-end^[5].

If Anthropic targets an October IPO at or near that valuation, it would eclipse SpaceX as the largest public offering in history^[5]. The valuation is not formally fixed within the company, and the timeline is still under discussion^[5].

OpenAI, which filed its own confidential prospectus shortly after Anthropic, presents a more uncertain picture. Reuters reported in June that OpenAI targeted up to a $1 trillion valuation with a potential September IPO^[6], but Morningstar notes that as of late June, OpenAI had not held pre-IPO meetings or set an official timeline, and some signals suggest executives may delay to 2027^[6]. An OpenAI delay would not reduce total supply — it would push it into next year — but it would give the market more time to absorb the Anthropic offering before facing another mega-listing.

The De-Equitisation Tailwind: Still a Tailwind, Less of One

Steve Sosnick at Interactive Brokers frames the structural shift well. For years, a combination of buybacks and takeovers shrank the supply of publicly traded shares, creating a self-reinforcing demand dynamic. That dynamic is now being tested from two directions: large IPOs adding share supply, and cash-rich companies like Alphabet and Meta raising capital rather than just returning it^[2].

None of this is cause for immediate alarm. Goldman found little evidence that investors are struggling to absorb additional supply — offering discounts and post-pricing performance show “no abnormal sign of indigestion”^[1]. The market absorbed a record $252 billion of issuance in Q2 without breaking^[1]. Sosnick himself acknowledges that corporate insiders will not sell indiscriminately into a reluctant market, as the SpaceX non-event demonstrated^[2].

But the direction of travel matters. If hyperscaler capex exceeds cash generation through 2027, if Anthropic and then OpenAI come to market at trillion-dollar-plus scales, if SpaceX’s staggered lockups eventually release selling pressure on the 12.9 billion shares still locked — the net supply of U.S. equity is moving from shrinking to growing, even if buybacks keep the aggregate balance positive. The tailwind becomes less of a tailwind.

Key Supply-Demand Variables to Track

What to Watch Next

1. SpaceX subsequent lockup tranches. The August 6 unlock passed without a fire sale, but the staggered schedule runs through mid-2027^[3]. Watch whether any large early backer files a Form 4 disclosure trimming its position, and whether SPCX approaches the $135 IPO price — the level at which the price-based early-release provision would activate and potentially triple the float^[3].

2. Anthropic’s IPO pricing and timing. Early investor meetings are high-level and valuation-free so far^[4]. The transition to financial roadshow discussions — where specific numbers get put on the table — will be the signal that an October listing is real rather than aspirational. Whether investors accept a $2 trillion valuation^[5] at a time when the market is absorbing record issuance will be the single most important test of demand depth this year.

3. Hyperscaler equity issuance. Alphabet’s $20 billion stock sale^[2] may be the first of several. If Meta or Amazon follow with secondary offerings to fund AI capex, the supply side of the equation grows faster than Goldman’s base case assumes.

4. OpenAI’s decision window. A September IPO would stack two mega-listings within weeks, testing whether the market can absorb both simultaneously^[6]. A delay to 2027 spreads the supply over a longer period — easier to digest, but building a larger overhang for next year.

5. Buyback execution vs. authorization. Record authorizations do not automatically translate into record execution. If companies begin channeling cash toward capex rather than repurchases — the pattern Alphabet and Meta have hinted at — the $1.4 trillion estimate could prove optimistic on the buyback side at the same time issuance runs above expectations.

The honest forecast: the supply-demand balance that quietly supported U.S. equities for years is loosening, not breaking. Buybacks of $1.4 trillion are a powerful offset, but they are running against a rising tide of AI-driven issuance, mega-IPO lockup releases, and hyperscaler capital needs that may exceed cash generation through 2028. The base case is that corporate demand stays larger than supply through year-end — call it 65/35 — but the gap is narrowing, and the composition is shifting in ways that make the market more sensitive to any single large deal that prices poorly. The next inflection point comes when Anthropic puts a number on its offering.