The hundredfold and the rails: Goldman Sachs is building a market the S&P won't let you into
On 4 June 2026, Goldman Sachs put a number on the future. The bank's analysts now project that SpaceX's artificial-intelligence revenue will grow a hundredfold by 2030 — a 9,900% expansion that, in dollar terms, lands somewhere north of $300 billion, according to a Financial Times report picked up by Reuters. The same week, the bank quietly launched a tokenized money fund with crypto custodians Apex and Archax. And the S&P 500 committee, per reporting the same day, will not be fast-tracking SpaceX into its benchmark index. The three moves are not separate stories. They are one story.
Goldman Sachs is doing what the public market will not: building the plumbing to capture AI's value while it remains in private hands. The "100x" projection is not analyst bravado. It is the marketing brochure for a parallel financial system — tokenized, on-chain, off-exchange — that the biggest balance sheet on Wall Street is now actively constructing because the official one keeps the door shut.
A number, and the assumption inside it
Start with the projection. A hundredfold revenue increase in four years, to roughly $322 billion, is not a forecast in any conventional sense. It is a bet that SpaceX's AI division will execute at a scale no private company in history has ever executed at, and that demand for whatever it sells will follow. Neither condition is implausible on present trajectories — but neither is supported by current revenue, either, which the available reporting does not break out at all. Goldman is pricing the upside of an asset that, by the S&P's own rules, retail cannot own.
That detail matters more than the headline figure. A 9,900% projection issued by the lead left-underwriter of SpaceX's private rounds is, in effect, a price target on an instrument the rest of the market cannot buy. It tells the people who already own SpaceX what they are sitting on. It tells everyone else what they are locked out of.
The benchmark that won't open
The same 24 hours brought the second data point. Per S&P Dow Jones Indices, SpaceX will not be fast-tracked into the S&P 500. The mechanics are dull: there is no public float, no ordinary share registration of the kind the index methodology requires, no liquidity test the company could plausibly meet. So the largest AI-adjacent private enterprise on the planet stays off the benchmark that defines what the average pension, the average target-date fund, and the average 401(k) are allowed to own.
This is not a complaint. It is a structural feature. The S&P 500 is a public-market instrument, and the public market is not where the AI capex is being deployed. Index investors — which is to say, almost everyone saving for retirement in the West — are, by construction, holders of yesterday's industrial composition. They own the railroads of 1957, updated quarterly. The railroads of 2026 are private, and the index committee is not in the business of rewriting the rulebook to include them.
The tokenized detour
So Goldman is building a different door. The tokenized real-estate fund, run on the bank's GS DAP platform with Apex Group providing fund services and Archax on the digital-securities side, is the first major deployment of a piece of infrastructure that has been in development for the better part of a decade. Read the announcements carefully: this is not a pilot. It is a production launch, with the bank as anchor and the crypto-native firms as plumbing.
Real estate is the right starting asset for a reason. It is a multi-trillion-dollar store of value, illiquid, and politically easy to defend as "modernisation" rather than speculation. Once the rails are in place — once KYC, custody, settlement, and reporting all work on-chain under a bank-controlled wrapper — every other private asset can be poured through the same pipe. The next one, presumably, will be exactly the kind of revenue stream Goldman is projecting for SpaceX. The fund is not the product. The platform is the product.
What this actually is
Strip the marketing and the picture is straightforward. The public market is becoming the venue for slow-growth incumbents. The AI economy is being built in private, with capital supplied by sovereign wealth funds, endowments, and a small circle of asset managers willing to write the cheques. The benchmark index will not adjust. And the biggest bank on the Street is now running the engineering work to make the private economy tradable — at scale, in tokenized form, on infrastructure it controls.
This is the part that deserves plain language. Coverage routinely frames tokenization as a democratizing force, a way to "open private markets" to smaller investors. The structural read is the opposite. A tokenized fund is a way to extend the reach of private-market returns to investors who are not yet accredited, by selling them a wrapper that depends on the same gatekeepers for pricing, access, and redemption. The gate does not come down. It is widened selectively, with a fee attached.
The stakes
The stakes are concrete and most of them land on the same people. Every pension fund, defined-contribution plan, and indexed retirement account in the Western savings system is, by design, a passive investor in the S&P 500. If the AI economy's largest operators stay private — and the benchmark index refuses to accommodate them — then the retirement-savings system is structurally short the most important productivity story of the decade. Tokenized funds are pitched as the remedy. But tokenized funds backed by tokenized real estate, settled on permissioned ledgers, and serviced by subsidiaries of the same banks that underwrote the 2008 mortgage complex, are not a deregulatory revolution. They are a re-aggregation. Capital is not escaping Wall Street. It is being rerouted through a new wing of the same building, with fewer disclosure obligations and a thinner public ledger.
The hundredfold and the rails
The hundredfold projection is the load-bearing number. If Goldman is right, then every index investor is missing it, and the case for a parallel private-market system is closed. If Goldman is wrong, the rails are still being laid, and the next time the projection lands — for the next private AI giant, the next carrier of the AI capex story — the architecture will already be in place to capture it. The 4 June announcements were not a coincidence of timing. They were sequencing. The forecast establishes the prize. The tokenized fund establishes the plumbing. The S&P's refusal establishes the demand.
Either way, the public market has not been invited. And the public-market investor — the one whose retirement is, by design, the average of whatever the index contains — has not been told what is being built in the room next door. That, more than the hundredfold, is the news.
Desk note: The wire treated these as three discrete stories — an analyst note, a product launch, and an index decision. Monexus treats them as one, because the public-market consequences only make sense when read together.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- http://reut.rs/3SnjWR9
- https://en.wikipedia.org/wiki/SpaceX
- https://en.wikipedia.org/wiki/S%26P_500
- https://en.wikipedia.org/wiki/Goldman_Sachs