arvy Notes · IPO Analysis

SpaceX at 90× Sales — Who Is Buying When the Smartest Investors Are Selling?

The IPO Class of 2026 will bring potentially several trillion US dollars of new tradable equity supply to the market. An analysis of the mechanics — and why the most attractive entry point is rarely the listing day.

Written by Thierry Borgeat, Co-Founder of arvy · Reviewed by Patrick Rissi, CFA and Florian Jauch, CFA · 26 May 2026 · 8 minute read

A few weeks ago, in "The IPO Class of 2026", we described why this wave of listings deserves Swiss investors' attention. With the SpaceX S-1 filing, that thesis has now materialised in a concrete data room. What is documented there is not an investment case — it is a textbook setup for asymmetric supply dynamics, and at the same time a discipline test for any investor who has held the broad US market over the past two years.

1. The Valuation: 90× Sales with a Net Loss

The headline numbers from the S-1 are unusual in their clarity:

The largest IPO in history — Saudi Aramco, the world's largest energy company with a demonstrably profitable core business — priced at 6× sales. SpaceX is asking for fifteen times that multiple, with a structural net loss attached. Even relative to the most ambitious growth names of today's Magnificent Seven, that is an extraordinary entry valuation.

We are not arguing that SpaceX is not a remarkable company. Starlink is a commercial reality with real customer utility, and Falcon 9 has fundamentally changed the economics of spaceflight. But between "remarkable company" and "90× sales with a loss at listing" historically lies several years of operational delivery. That delivery the investor is being asked to fully prepay on day one.

2. The TAM Is 93% Imagination

Source: SpaceX S-1 Filing, May 2026. "Estimated TAM by Segment" — the company's official market sizing. Reading note: three buckets (Space $370B, Connectivity $1.6T, AI $26.5T) aggregate to the disclosed TAM of $28.5 trillion.

The company quantifies its Total Addressable Market in the S-1 at USD 28.5 trillion. That number is larger than the GDP of the United States. The breakdown deserves a very careful read:

The actual core of the company — rockets, satellites, Starlink connectivity — represents in this self-presentation less than 7% of the claimed market. Over 93% of the TAM is AI software that SpaceX does not build, does not sell, and does not own. The company is at its core a rocket and satellite business with one major AI compute contract — which we will come to next.

Every IPO inflates its TAM. That is an established discipline of market communication and not a new phenomenon. But SpaceX has set a new record in that discipline: 79.6% of the cited market opportunity sits in a single line item called "Enterprise Applications" — that is, enterprise software. A market in which SpaceX today owns no product, no sales organisation, and no demonstrated competence.

3. The Largest Customer Is the Most Fragile

The S-1 documents, in its later pages, a contract that on its own changes the entire investment thesis:

That single contract approaches the entirety of SpaceX's prior-year revenue. It is also legally cancellable on three months' notice — by either side. It is exactly the kind of risk factor that gets buried on page 47 of the prospectus and becomes the entire story two earnings calls later.

We do not want to overstate this: Anthropic is a well-funded company, and the underlying compute demand is real. But the structural asymmetry — a single customer whose contract volume exceeds the entire established core business, with 90-day termination — is precisely the type of concentration risk that typically does not get appropriately priced at the listing date.

4. The Future Markets Read Like Science Fiction

Source: SpaceX S-1 Filing, May 2026, "Future Markets" section. Verbatim reproduction of the future market segments listed by the company.

In the section titled "Future Markets", the S-1 lists verbatim: point-to-point terrestrial travel, space tourism, in-orbit manufacturing, passenger and cargo transport to the Moon and Mars, energy production on the Moon and Mars, manufacturing capabilities on the Moon and Mars, asteroid mining.

This is not an investor letter from the communications team — it is a mandatory filing with the SEC. When a company is asking 90× sales today on the assumption that asteroid mining will eventually fund the cash flows, it isn't selling a security. It is selling a worldview.

That worldview may well turn out to be correct. The operating question is not whether Elon Musk and his team are exceptional engineers (they are), but whether a Swiss investor in May 2026 should prepay the asteroid-mining phase at full price.

5. The Smartest Investors in the World Are on the Selling Side

A look at the cap table makes what may be the most important point:

These investors do not sell their best positions cheaply. When the seller is Peter Thiel, Mike Moritz, Marc Andreessen, a Saudi sovereign wealth fund, and the wealthiest software CEO in the world — the more useful question is not "who is selling?" but "who is buying, and why are those buyers convinced they are paying a better price than those sellers received?"

For historical context: Amazon IPO'd at USD 438 million. Google at USD 23 billion. Both became trillion-dollar companies. The IPO Class of 2026 will likely produce one or two comparable successes. But the best time to buy them was historically almost never the listing day. The honest price typically appears later, after the lock-ups have expired and the market has had time to decide what the business is actually worth.

When the smartest money in the world is the seller, the better question is who is on the buying side.

6. The Real Mechanic — the Supply Wave of the IPO Class 2026

SpaceX is the most prominent, but not the only, element of the 2026 wave. Together with the expected listings of OpenAI and Anthropic, potentially several trillion US dollars of new tradable equity supply is coming to market — all driven by the same AI narrative that, by our classification, already dominates around 60% of the S&P 500.

The mechanic we are seeing here is not speculative — it is structural:

Phase 1 — Listing (Day 0 to ~Day 90)

Lock-up periods typically bind insiders and pre-IPO investors for around 90 days. During this phase, the actually tradable free float is small, which can push prices higher. Retail enthusiasm meets constrained supply.

Phase 2 — Index Inclusion

Once market capitalisation and trading history meet the index criteria, the new names are mechanically added to the Nasdaq-100, total-market, and eventually S&P 500 indices. Passive funds — meaning the majority of global index capital — must buy. That is rule-bound, not conviction-driven.

Phase 3 — Lock-up Expiry

In the very window in which the passive buying kicks in, the lock-up period expires. Insiders, early VC investors, and employees with vested equity become free to sell. Mechanically, the setup that we view as the central risk factor materialises:

Passive ETF buying after index inclusion provides the exit liquidity for legacy shareholders — precisely when those shareholders want to sell.

This is not a conspiracy and not malice. It is how the system mechanically works: index rules require purchases, lock-up expiry permits sales, and the two collide in a window. Anyone entering at the wrong moment is structurally buying the inventory that the smartest early investors are unloading.

7. What This Means for Other Tech Names

The supply wave has secondary effects that we factor into positioning. Three observations:

Tesla — negative in the short term. A portion of the "Musk-exposure" capital is likely to rotate out of TSLA into SpaceX. The indirect SpaceX stake on Tesla's balance sheet is too small relative to market capitalisation to offset that rotation.

NVIDIA — neutral to slightly positive. SpaceX is an important GPU customer, and fresh capital in the AI compute ecosystem structurally supports demand. In the short term, IPO hype can pull attention and speculative flows away from NVDA.

Broad market — temporary liquidity drain, then passive buying. Highly valued tech names can serve as a funding source in the pre-IPO phase. After index inclusion, passive buying takes over — primarily benefiting legacy shareholders.

Important: we expect this to remain primarily a tech- and sympathy-isolated phenomenon — episodic pressure in the hype segment, not broad market stress. The forthcoming IPOs (SpaceX, OpenAI, Anthropic) are all driven by the same AI hype narrative and need to deliver operationally "big time" to justify their valuations. Disappointments would primarily hit the AI and growth cluster, while the rest of the market should remain comparatively robust.

8. The Rate Component — the Overlooked Variable

A second concern is real interest rates, which remain elevated. Real rates are the central valuation metric for growth equities — they determine how the present value of cash flows projected far into the future is calculated today. That is precisely the segment of rate-sensitive long-duration equities at the centre of the 2026 IPO class.

The lesson from 2021 should not be forgotten: at that time, euphoric IPO valuations met a rate inflection point — with correspondingly painful consequences for new buyers. The macro context today is not identical, but the structural vulnerability of the rate-sensitive IPO class to a hawkish rate surprise is the same.

Implications for Investors — arvy's Positioning

At arvy, we are positioning defensively around this IPO wave. That has three concrete implications for our Quality equity fund and our savings-plan allocations:

First — no pre-IPO or day-one allocation to the 2026 class. Our investment process requires established cash-flow histories, demonstrated profitability, and valuations in historical context. None of the three prominent candidates meet those criteria at the listing date. That is not a statement about company quality — it is a statement about the phase of the investment cycle in which we invest.

Second — heightened vigilance on sympathy-driven high valuations in the existing portfolio. The AI-narrative-driven component of the market typically gets further fuelled in phases of IPO euphoria. We use such phases for discipline, not for add-ons.

Third — patience as strategy. Medium- to long-term, we remain constructive on the broader AI compute and Musk ecosystem. But the historical data is unambiguous: the more attractive entry point typically lies 6–24 months after the IPO, once hype, lock-up waves, and insider selling have worked through and the honest price has formed.

We are a Swiss asset manager, FINMA-regulated under CISA Art. 24, with our own capital invested in the same fund as our clients. Our task in phases like this one is not to ride the next wave — it is to keep our clients in the position to stand, in 18 to 24 months, as buyers where today the insiders stand as sellers.

The allocation logic we described in the Quality note only gains weight through this IPO wave: if 60% of the broad index is already driven by a single narrative and the next two to three trillion USD of new supply originate from exactly the same narrative, then a Quality pillar with low AI-narrative exposure is not a speculative bet. It is the structural answer to a concentration and supply asymmetry that has few historical parallels.

Vanguard Advisor Alpha estimates the behaviour-gap cost of the average investor at 1.5% per year. The majority of that cost is incurred in phases like the current one — when a listing day becomes a cultural obligation and discipline is most uncomfortable.