Socializing the Upside: The Record SpaceX IPO and the Case for a Social Wealth Fund

On 2026-06-12, SpaceX started trading on the Nasdaq as SPCX. The company sold 555.6M shares at $135 and raised $75B, with an $11.2B over-allotment option on top. That put the implied valuation at $1.77T, about 2.9× the size of the previous record holder, Saudi Aramco (2019), which raised $25.6B. The stock opened at $150, ran up to $168.75, and closed at $160.95, a gain of +19.2% on the day. No company had ever gone public at anything close to this scale.

This piece does two things. The first is to read the deal the way an investor would, working through the valuation, the risks, and the market mechanics with no policy agenda attached. The second is to use the IPO as a way into a question about public economics. When the government underwrites a private company through bailouts, subsidies, and procurement, and that company turns into something worth $1.77T, who gets the upside? The argument here is that the public should hold a real and lasting stake in it, pooled into a Social Wealth Fund of the kind Norway and Alaska already run. You can take the investment analysis and leave the policy, or the other way around. Each half stands on its own.

---

I · The Event

Is the largest IPO in history priced like a business or like a moment? And what does an offering this big do to the market around it? The historical record of mega-IPOs has a lot to say about where SPCX goes over the lockup period, the index-inclusion window, and the next one to three years, so that is where we start.

The deal and its mechanics

The structure is simple enough. SpaceX raised $75B by selling 555.6M shares at $135, with an over-allotment option, the “greenshoe,” covering another $11.2B. The greenshoe is a stabilization tool. Underwriters sell a bit more stock than the deal officially contains, and then they either buy shares back in the market if the price sags, which props it up, or exercise the option to cover themselves if the price climbs. At $1.77T, SpaceX came public as roughly the seventh-largest company in the United States, ahead of Tesla.

This is the biggest IPO ever in plain dollar terms, and it stays the biggest even after you adjust older deals for inflation. The league table below makes the comparison in 2026 dollars. SpaceX clears Saudi Aramco (2019) and the big late-1990s telecom listings in real terms.

Two facts about the structure end up driving everything else. The first is control. SpaceX has a dual-class share structure, so the founder keeps voting power well beyond his economic ownership. Buy the stock and you are buying a claim on the cash flows, with no real say in how the company is run. The second is scarcity. The piece of the company that actually trades is tiny. At the $1.77T valuation and the $135 price, the public float works out to only about 4.9% of the shares outstanding. A company worth almost $1.8 trillion with that little stock in public hands is going to be volatile, and it will distort any index it enters. Both themes come back later.

Valuation

SpaceX booked $18.7B in revenue for FY2025 and lost $4.9B on a GAAP basis. With no profits there is no price-to-earnings ratio to quote, so the $1.77T valuation works out to about 95× trailing sales. On its own that number does not tell you much, because it lumps together three businesses that have almost nothing in common. The honest way to value the company is to take them one at a time.

That approach is called sum-of-the-parts. You split the company into its real businesses, value each one against comparable firms, and add the pieces back up. SpaceX has three. Starlink is the part that makes money. It pulls in $11.4B of revenue, growing around 50% a year, with $4.4B of operating income and 9M+ subscribers, and it has room to run in direct-to-cell service and global broadband. Launch, meaning Falcon and Starship, is a dominant franchise that throws off cash, and its cost per kilogram to orbit keeps falling. Then there is xAI, the odd one out. It arrived through the early-2026 merger, it loses money, and it is a bet on frontier AI rather than a going concern, with an $6.35B operating loss against $12.7B of capital spending. Pricing it means picking a number off private AI valuations and accepting a wide band around it.

The builder below lets you set the revenue multiple on each segment and watch the total move. A defensible base case comes out around $529B. Even a generous bull case only reaches $1.01T. Either way, the $1.77T the market is paying sits above the top of any reasonable sum-of-the-parts range.

There is another way to look at the same gap. A discounted cash flow model takes a company’s expected future cash and translates it into a value today, shrinking each future dollar a little because money now is worth more than money later. Run that model backwards and you can ask what the current price already assumes. For SpaceX, solving for the revenue growth rate that would justify a $1.77T enterprise value, at a 20% long-run profit margin and a 11% discount rate, gives you about 56% revenue growth every year for a decade. That is the kind of growth that would make SpaceX one of the largest companies on earth by revenue sometime in the mid-2030s. It could happen. But it is the bet you are making at today’s price, whether or not you think of it that way. The grid below shows how the required growth shifts as you change the margin and the discount rate.

You can also just lay out the cases. Bull, base, and bear, each with its own story and a probability attached. The resulting valuations are all over the map, and weighting them by probability gives a DCF value of around $718B against a $1.77T market price. The distance between those two numbers is what the market is paying for the story. The bull case can get there, if Starlink becomes the world’s connectivity layer, Starship collapses the cost of launch, and xAI turns into a top lab. The base and bear cases cannot. Owning SPCX at this price means betting that the bull case is the likely outcome rather than the lucky one.

First-day pop and underpricing

The +19.2% jump on day one handed about $14.42B of value from SpaceX to the investors who got IPO allocations. That is money “left on the table,” the difference between what the shares sold for and where they closed. This happens with almost every IPO. Banks underprice new issues partly to reward the big investors who help set the price during the roadshow and partly to keep favored clients happy.

The interesting thing is how unremarkable the pop was. Line up the first-day returns of past mega-IPOs and SpaceX lands at the 53rd percentile, behind Snowflake, Airbnb, Reddit, and Alibaba, and only just ahead of the +17% median. The largest offering in history had an average first day. That tells you something. Demand was real and the float was small, but the bankers priced it close enough to the moment’s fair value that it never went vertical the way scarcer, smaller deals sometimes do.

The longer view is less flattering. The drift chart below lines every comparable up at its own IPO date and follows the cumulative return from there, so you can compare companies that went public years apart on the same clock. The pattern is split, and on average it runs negative over one to three years. This is Jay Ritter’s long-run underperformance result, and you can see it directly. A few winners like Arm and Reddit sit against real disasters like Rivian, Coupang, and Lucid.

Risk analysis

The risk register scores 16 risks by how likely each one is and how much damage it would do, then plots them as a heatmap you can filter. Governance sits at the top, between the founder’s super-voting control and the dependence on one person. Right behind it is valuation. A price that needs 56% growth for a decade has no slack if the AI and space story loses momentum. Then come the financials, where the $4.9B loss is mostly xAI burning cash. One more risk, the heavy reliance on US government revenue, also happens to be the hinge of Part II.

Market structure, flows, and price discovery

Scarcity is the structural story here. With only about 4.9% of the shares trading, three forces will set the price in the near term.

• The lockup waterfall. After an IPO, insiders cannot sell right away. Their shares unlock in stages. A slice typically frees up at 90 days, a much bigger tranche at the 180-day cliff, then the employee and RSU shares, and finally the founder’s overhang. The 180-day cliff is the single largest of these, somewhere around $703B of stock landing on a market that trades thin daily volume. When a lot of newly sellable stock hits at once, the price tends to drop, and the schedule is known in advance, so this is a supply shock you can watch coming.
• Index inclusion. SpaceX could qualify for the Nasdaq-100 quickly, which would force every fund tracking that index to buy. The S&P 500 is harder, because it requires four straight quarters of GAAP profit, and the $4.9B loss disqualifies the company for now. So the bigger wave of passive buying probably waits until SpaceX strings together real profits. The dashboard sizes how much each inclusion would pull in.

Broader-market and sector spillovers

Does a near-$1.8-trillion listing drag the stocks around it? The peer event study looks for that effect in listed space, satellite, and defense names, plus Tesla as a proxy for the wider Musk orbit. So far there is only the debut session to work with, so the numbers are thin, and the pipeline re-runs them every week as more days pile up.

The method is standard. You take a stretch of trading before the IPO and fit a simple relationship between each peer and the overall market. Then, in the window around the debut, you check how each stock actually did against what that relationship predicted from the market’s move alone. Whatever is left over is the abnormal return, the part you can plausibly tie to the event.

Pulling it together

The base case is that SPCX is a very good company carrying a very demanding price. The sum-of-the-parts ($529B) and the probability-weighted DCF ($718B) both land well under the $1.77T mark. The reverse DCF wants 56% growth a year for ten years. The first-day pop was ordinary. The mega-IPO track record runs negative over one to three years. And a wall of locked-up stock is scheduled to come loose.

None of that is a prediction that the stock falls. Expensive stories can stay expensive for a long time, and the bull case really is possible. The point is narrower than that. At this price the downside outweighs the upside, and that is the honest read.

Predictions

To keep all of this honest, four specific predictions are written down and dated at publication, then scored every week against what actually happens. They have bands around them, and they will be marked right or wrong out in the open.

1. Three-month return against the Nasdaq-100: mild underperformance.
2. Twelve-month return against the Nasdaq-100: underperformance, on the valuation gap and the lockup supply.
3. The 180-day lockup: a negative abnormal return around the cliff.
4. One-month peer spillover: a small positive read-through for listed space names.

The tracker shows the live price together with the scenario DCF fair-value lines, which are reference levels rather than targets, and the stock currently trades above even the bull line. The scorecard sits underneath. As of day 1, SPCX is at $160.95, +19.2% against the $135 IPO price, and all four calls are still open.

Strongest objections to Part I. First, that a sum-of-the-parts built on public-market multiples cannot capture a private-quality monopoly. That is fair. Starlink and launch may deserve scarcity premiums no listed comparable shows, which is exactly how the bull case reaches the market price. The argument was never that the price is impossible, only that it treats the bull case as the expected one. Second, that the reverse DCF depends on its inputs. True, which is why it is shown as a whole grid instead of one number, and across the believable range of margins and discount rates the required growth stays enormous. Third, that mega-IPO base rates do not apply to a company that defines its category. Maybe so. Survivorship runs both ways, and the comparison set was picked to include the flops alongside the Arms and the Reddits.

What the data cannot see (Part I). The private segments are disclosed at low resolution, and the xAI figure is a range rather than a point. The exact float and lockup calendar are modeled from typical deal terms until the filed prospectus can be read line by line. The event study describes; it does not prove causation, and it cannot pull the SPCX debut apart from everything else moving that week. And the predictions are judgment calls with bands, not confidence intervals from a model.

---

II · The Entanglement

How much of the value that got capitalized at the IPO was really underwritten by the public, through contracts, subsidies, guarantees, and spectrum? And if past bailouts had gone differently, if the government had held onto the equity it took instead of selling it back, what would that be worth now?

SpaceX’s public scaffolding

Pull SpaceX’s awards out of the federal spending database and you get $15B in identified contract obligations. Most of it runs through two channels. NASA paid for Commercial Crew, Commercial Resupply, and the Artemis lander. The Department of Defense paid for national-security launch through the Space Force. The chart below stacks it up over time.

A point worth making clearly. These contracts are payment for work, not handouts. SpaceX flew the cargo, carried the astronauts, and delivered the capability, usually for less than anyone else was charging. Nobody handed the company money for nothing. The argument is more specific than that. The public took on the early risk, by being the only customer when there was no commercial market, by funding the hard parts of building the capability, and by handing over spectrum and launch access. When that risk paid off and turned into a $1.77T company, the public got none of the equity. Add the $885.5M FCC broadband award that was granted in 2020 and later pulled, plus some state and local incentives, and the full ledger looks like this.

What if we’d kept it?

The clearest example of giving up the upside is not SpaceX at all. It is 2008. During the financial crisis the government took real equity and warrants in exchange for the rescue money, around 92% of AIG, about 60% of General Motors, big positions in Citigroup and across the banking system. And then it sold all of it back between 2009 and 2014, as soon as that was politically comfortable.

The usual verdict is that TARP turned a profit, and on a cash basis that holds up. The government put out about $397B across the major programs and got back about $424B, a gain of +$27B. But that is the answer to the wrong question. The better question is what the public would have now if it had kept the money working instead of handing it back to general revenue. Take each program’s proceeds, reinvest them in a broad market index from the day they were sold through today, and the retained portfolio would be worth about $2.30T. The gap, roughly $1.88T, is what selling early cost. One of the longest bull markets on record came right after the crash, and the public, having eaten the downside, was no longer in the room for the recovery.

This is an illustration rather than a causal estimate, and the limits are worth stating. It assumes the stakes could have been held and indexed. It sets aside the chance that some of these firms would have collapsed without active public restructuring. And it uses a price index that actually understates total return by leaving out dividends. Even so, the direction is obvious, and the size is big enough to survive a heavy discount. The slider in the dashboard lets you impose as much underperformance as you like and watch how much it takes to close the gap. It takes a lot.

The bridge to a rule

Now run the same logic forward over the kind of public support SpaceX received. Suppose a warrant-priced equity condition had been attached to the subsidy-like portion of that support, not to the arm’s-length launch contracts but to the risk the public was absorbing underneath them, plus the outright subsidies. On conservative assumptions the fund would hold something like $12B in SpaceX today, from that one company alone. Every input is a slider in the dashboard, and they are all set deliberately low.

Strongest objections (Part II). Contracts are payment, not gifts. Correct, and the analysis keeps them separate, putting the equity logic only on the subsidy-like slice and the outright support, never on a competitively bid launch. Government equity gets in the way of restructuring. A real worry, which Part III handles with non-voting, arm’s-length stakes. This is all hindsight. We know the market came back, and the counterfactual is illustrative precisely because nobody could have known in 2009, which is the case for a standing rule that runs through every cycle rather than a one-time market-timing call. Retroactive application raises legal problems. Flagged here, dealt with in Part III.

What the data cannot see (Part II). USAspending shows identified federal awards. It does not show the value of spectrum grants, regulatory forbearance, or the quiet advantage of being the government’s anchor customer. The TARP counterfactual cannot see the world where the stakes were kept, where bank behavior, dividends, and political pressure would all have gone differently. These are rough magnitudes meant to motivate a mechanism, not precise estimates of something knowable.

III · The Fund

If the public is going to underwrite private companies through bailouts, subsidies, and procurement, how does it actually hold a stake in the upside instead of giving it away? How does it pool that stake into a fund that pays everyone a dividend, and do it without the state ending up running companies badly?

The proposal is a Social Wealth Fund fed from several sources, and the equity condition on public support is the one the SpaceX story points to most directly. I will spell out the equity condition first, then the wider set of inflows a real fund would lean on, then the distortion question that hangs over all of them. None of it relies on the tax in Part IV. The two fit together, but they are separate policies, and each can stand alone.

It already exists, six times over

A Social Wealth Fund is not hypothetical. Norway’s Government Pension Fund Global owns something like 1.5% of every listed company in the world and spends only about 3% of itself a year. Alaska has cut every resident an annual check from oil money since 1982. Singapore runs Temasek and GIC, Australia has its Future Fund, and Saudi Arabia has the PIF. They carry different mandates, and they teach different lessons.

The lessons are not subtle, and they shape the design. Norway shows the value of professional management at arm’s length, of indexing broadly instead of trying to pick winners, of an ethics council and real transparency, and of a spending rule strict enough to keep the fund alive forever. Alaska shows that a visible check to every resident is what makes a fund impossible to kill, because people defend the thing that pays them. Saudi Arabia’s PIF shows the other side, what happens when a fund gets concentrated and political.

The equity condition

Here is the rule. When a company takes a bailout, or gets subsidy or procurement support above some threshold, the public takes a matching equity stake, and that stake goes into the fund and stays there. The details work like this.

• Trigger and threshold. Direct bailouts and capital injections. Loan guarantees, priced as the options they really are. Grants and subsidies over a set size. Large sole-source or cost-plus contracts. Tax breaks above a threshold. The line runs between support that should turn into equity, like a bailout or a subsidy or a guarantee, and ordinary payment for a service, like a launch the company won in open competition, where taking equity is much harder to defend.
• Sizing the stake. A transparent, warrant-style formula. The stake is worth some multiple of the support’s fair value, set at the date the support is given. This is what the 1979 Chrysler rescue and the TARP bank programs already did. The public gets paid like any other investor putting capital at risk.
• Form of the equity. Non-voting shares, or shares voted by an independent steward, as the default. That takes the “government will run the company” problem off the table while keeping the financial claim, the same split between cash flow and control that SpaceX itself imposes on its public shareholders.
• Prospective first, retroactive with care. The clean version applies to new support going forward. Retroactivity shows up two ways here, as the illustrative counterfactual in Part II, and as a defensible condition of continued eligibility. Keep taking public contracts and subsidies if you like, but new support now comes with the condition attached. Nothing already granted gets seized, which sidesteps the worst of the takings problem.

One mechanism among several

Tie the fund only to the companies that get bailouts and subsidies and you end up with a badly balanced portfolio. Those companies bunch up in a few corners of the economy, finance in 2008, autos, aerospace and defense, energy. A fund that is supposed to pay everyone a steady dividend wants the opposite of that. It wants to own a little of everything, the way Norway’s fund holds a sliver of nearly every public company in the world and so rides the average instead of betting on any one name.

So the equity condition is really just one inflow among several, and the more interesting question is the whole menu. Matt Bruenig’s Social Wealth Fund for America lays out most of it.

• Leveraged purchases. The fund borrows at the government’s low rate, buys a broad basket of higher-returning stocks, and pockets the spread. Alaska already does a version of this. Its fund is not stuffed with oil stocks; it holds a diversified global portfolio that oil money seeded.
• Countercyclical purchases. This is the “kept it” idea from Part II turned into a standing policy. When markets crash and the government is on the hook to steady them anyway, the fund buys stocks and holds them instead of handing the rebound back to private buyers. In 2008 the public ate the loss and skipped the recovery. A permanent fund would do the reverse, and soften the crash on the way.
• A share levy. Big profitable companies hand the fund a small number of newly issued shares each year, a fraction of a percent, instead of cutting a check. Because it is paid in stock, it never touches the company’s cash, and it does nothing to the decision to invest. Over time it spreads public ownership across the whole corporate sector, which diversifies on its own. This was the heart of Sweden’s Meidner plan, rebuilt around a citizen’s dividend rather than union control.
• Estate and wealth taxes paid in stock. Most large fortunes are inherited rather than earned, and an estate is a natural moment to take a cut. When the asset is stock, the bill can be settled in shares, the same in-kind move from Part IV, which drops straight into the fund without anyone being forced to sell.
• Public-asset leases. The public already owns the airwaves, big mineral and land rights, and the like. Send a share of the leasing revenue into the fund and a string of one-off auctions becomes a permanent endowment.
• Financial-transaction and carbon taxes. Small, broad levies that do double duty, trimming pointless high-frequency churn or pricing emissions, and feeding the fund with what they raise.

A few of these obviously rhyme with the rest of the piece. Countercyclical buying is the 2008 lesson written into law. Paying a tax in stock is the same mechanism that solves the liquidity problem in Part IV. And because most of them pull from the whole economy rather than a short list of rescued firms, they build in the diversification a citizen’s dividend needs. The equity condition then sits on top as a targeted layer, catching the upside in the specific cases where the public carried real, identifiable risk, which is the case SpaceX makes so vivid. No single one of these does the job by itself. The point is that there is a lot of room to design here, more than the debate usually admits, and a serious fund would use several at once.

There’s plenty of precedent

Taking equity in exchange for public support is a normal thing to do. The 1979 Chrysler loan guarantee came with warrants, and Treasury made money on them. TARP took warrants and equity throughout the banking system, and the bank programs came out ahead. Britain ended up owning most of RBS and Lloyds. None of that was exotic. The new part of this proposal is simply keeping the stake and pooling it, rather than selling it back the first chance you get, which, as Part II showed, is what the public has always done.

Market distortion, deadweight loss, and the second best

The usual objection to any tax or public equity stake is that it gums up the market and creates deadweight loss. It puts a wedge between price and value, discourages whatever it touches, and shrinks the pie by more than it collects. That missing output, the part nobody gets, not the taxpayer and not the Treasury, is the deadweight loss, the little triangle in the textbook diagram. It is a real cost, and it deserves a real answer rather than a wave of the hand.

There are two answers. One is to put the cost next to the benefits and see which is bigger. The other is the theory of the second best, which changes what the cost even is.

The theory of the second best comes from Richard Lipsey and Kelvin Lancaster in 1956, and it makes one sharp point. In an economy that is competitive and clean everywhere, any new tax or wedge can only make things worse. That is where the deadweight-loss intuition comes from. But it is a statement about a perfect world, and it only holds in a perfect world. Once a bunch of markets are already distorted, and in any real economy they are, the result falls apart. A well-chosen new distortion can push the economy toward efficiency instead of away from it. “Never add a wedge” is good advice in a world that does not exist. The real question is not whether the fund’s mechanisms are perfectly clean, since nothing is, but whether they offset bigger distortions that are already sitting there.

And those distortions are specific.

• The bailout guarantee nobody pays for. Big banks and strategically essential firms already operate with an implicit promise of rescue. That promise is worth a lot. It lowers their borrowing costs, and they pay nothing for it, which quietly subsidizes size and risk-taking. This is a live distortion right now. An equity condition puts a price on it, so the public finally gets paid for the insurance it is already writing. That fixes a distortion rather than adding one.
• The realization loophole. As Part IV gets into, taxing only realized gains already pushes people to never sell, locking capital into concentrated positions for tax reasons instead of good ones. Accrual taxation, or paying in stock, loosens that grip.
• Rents and market power. A company with the scale, the network effects, and the public spectrum behind something like Starlink earns more than a competitive market would let it. Economists call the extra economic rent. Taxing pure rent is, in theory, free of distortion, because rent is infra-marginal. The holder collects it whether or not it changes the decision to produce one more unit. That is the old Henry George point, that you can tax rent without shrinking output, and it applies neatly to a stake in rent-heavy, monopoly-ish assets.
• Concentration bending policy. Concentrated wealth buys influence, and that influence creates more distortions down the line, the carve-outs, the captured regulators, the barriers that protect whoever is already on top. A fund that thins out that concentration goes after the source of those second-round problems.

Stack those corrections up and even a design that leaves some deadweight loss behind can still come out ahead, because the benefits are concrete too. Most Americans own little or no stock, while capital has been taking a bigger slice of national income than labor for years. A broad dividend hands the median household a claim on exactly the returns it is shut out of now, and a pure deadweight-loss tally never counts that as the gain it is. Countercyclical buying steadies the economy in a downturn, which is worth real money. And the whole design leans the same way, non-voting stakes, broad indexing instead of steering companies, a focus on rents and on risk that is already subsidized, payment in stock rather than forced sales, all of it chosen to add as little friction as possible while clearing away friction that already exists.

None of this proves the fund comes out ahead on efficiency. That depends on how big the existing distortions really are and on how well the thing is built, and those are empirical questions. What the second-best argument does settle is smaller but worth saying. The reflex that any policy like this must create deadweight loss and therefore must make us worse off is simply wrong. In a distorted world the clean-world intuition is not a reliable guide, and the case has to be made on the actual numbers instead of waved away with a diagram.

How big could it get?

The simulator takes inflows from the equity condition, grows them at a return in line with Norway’s long-run record, applies a spending rule, and traces out the fund’s size and the dividend it could pay. Put in $60B a year, compound at a 5% real return, pay out 3%, and after 30 years the fund is worth about $3.04T and pays roughly $364 to every adult, every year. Drop a 2008-style 35% crash on it at year 15 and it still ends up around $2.55T. The inflow is not a stretch. It is a small slice of what the government already spends on procurement and subsidies every year.

That dividend is small next to Alaska’s per-resident payout, but Alaska runs on one state’s oil, and a national fund drawing on the whole economy’s public-support flows is playing a different game. The broader point is just that a lasting, broadly held claim on the upside the public is already financing can actually be built, using numbers borrowed from funds that already work.

Strongest objections (Part III). Moral hazard, that guaranteed public equity might make companies chase subsidies. If anything it cuts the other way, because pricing the stake takes the shine off capture. The public now shares in whatever windfall the subsidy produces. State ownership distorts markets. Handled by non-voting shares, arm’s-length management, and broad indexing rather than steering firms, and more fully by the second-best argument above, where the mechanisms are built to clean up existing distortions, the unpriced bailout guarantee, the realization lock-in, untaxed rents, instead of piling on new ones. Politicians raiding the fund. Handled by a statutory or constitutional lock and a spending rule, the things that have kept Norway’s fund whole. WTO subsidy rules. The condition is a quid pro quo for support rather than a new subsidy, and it falls on every recipient alike. Retroactive takings. Handled by leading with the prospective rule and the continued-eligibility version. The Meidner plan died politically. Which is the whole reason this design puts a visible universal dividend at the center, the Alaska lesson, instead of slow, dilutive collectivization.

What the data cannot see (Part III). The simulator projects, it does not forecast. Real returns wander, the inflows depend on whatever political deal gets struck, and a run of bad years early would bend the whole path. It also cannot price the general-equilibrium effects on investment and company behavior, the thing the objections keep circling, where honest economists land in different places.

IV · The Tax

Most really large fortunes are mostly unrealized capital gains, and the income tax, built the way it is, barely sees them. The question for this part is whether you can tax gains as they accrue in a way that actually works. It has to be administrable. It cannot demand cash people do not have. It has to refund losses as well as tax gains. And it needs a fighting chance of surviving in court.

One thing up front. An accrual tax on unrealized gains is not the main way to fund a Social Wealth Fund here, and it is not being sold as settled policy. It is worth working through because it goes at something the equity condition in Part III does not touch, the way concentrated wealth can compound forever without ever being taxed or even disturbed. Whether a tax like this ends up doing more good than harm is an open question, and this piece does not pretend to close it. What it does is lay out how the mechanism works, how big the base is, and which design choices would make it defensible.

Buy, borrow, die

The trick the largest fortunes use to dodge the income tax is not complicated. You hold an asset that keeps going up and you never sell it. You borrow against it instead, at low rates, to pay for your life, and because a loan is not income, none of that is taxed. The asset is the collateral, and these are called securities-backed lines of credit. Then you die, and your heir’s cost basis resets to the current market value under IRC §1014, so the entire gain that built up over your lifetime simply vanishes for tax purposes. The income tax only fires when you realize a gain, and the very rich just never have to realize. The example below makes it concrete. On $1,000 left to compound for twenty years, buy-borrow-die pays nothing in income tax on the gain, while an accrual tax would have collected a real amount along the way.

The scale

This is not a rounding error. American households are sitting on something like $56T in unrealized capital gains, and the top 1% holds about $24T of that, roughly 43% of the whole. As the tax code is written, almost none of it ever shows up in the income tax base.

Two design problems

Two hard problems stand between the idea and a workable tax.

The first is liquidity, the “dry tax” problem, where you owe tax on a gain but have no cash in hand to pay it. The fix is to treat assets differently depending on what they are. Publicly traded stock can be marked to market every year, and selling a sliver to cover the bill is easy. Private or illiquid assets, where annual valuation is a nightmare, get two other routes. One is a deferral charge, an idea from the economists Alan Auerbach and David Bradford. You skip the yearly valuation entirely and instead pay an interest factor when you finally sell, calibrated so it works out as if you had been paying accrual all along. The other is to let people pay in kind, handing over shares instead of cash, which settles the bill without forcing a sale. Paying in kind kills the liquidity objection outright, and if the Part III fund exists those shares have somewhere natural to go, though the tax works fine either way.

The second is symmetry, and it is what separates a fair accrual tax from a confiscatory one. A tax that grabs your gains but ignores your losses is a one-way bet against you, and it badly overcharges anything volatile. So the design runs both directions. When a marked asset falls, you get a refundable credit for the loss, or you carry it back with the same interest factor. The chart below runs a bumpy six-year path. The one-sided version takes $29 per $100, while the symmetric version takes only $12. The government ends up as a real partner in the bad years as well as the good ones, instead of showing up only when there is money to take.

Revenue

So what would it raise? The simulator takes the taxable stock of unrealized gains and multiplies it by an expected return, a rate, and then the two things that really move the answer, how much avoidance and migration the tax triggers and how much the loss refunds cost. At the $1-billion-net-worth threshold, the design behind Senator Wyden’s Billionaires Income Tax, with a 25% rate and middle-of-the-road assumptions, it brings in something like $129B a year, or about $1.29T over a decade. The high estimates ride almost entirely on the avoidance number, which is why that number is a slider you can move yourself rather than an assumption hidden in a footnote.

What the tax is really for

The standard objection to any wealth or accrual tax is capital flight. Tax the rich and they leave, and the investment goes with them. The worry deserves a more careful look than it usually gets.

Start with the fact that public stock cannot actually pick up and leave. An American holding shares in an American company cannot move those shares to a tax haven to escape a US tax; the bill follows the person, not the certificate. Flight is a real issue for truly mobile things, some kinds of financial activity, and much less of one for concentrated public equity held by US residents, which is what this tax is aimed at.

Then there is the evidence, which is not kind to the flight story. Study after study finds that the wealthy move in response to high top rates far less than people expect. Professional networks, family, the legal system they understand, the houses they live in, all of it creates friction the simple models leave out.

And the lock-in argument runs both ways. Critics are right that an accrual tax takes away the ability to defer forever. But the system we have now is itself a distortion. Wealthy investors sit on concentrated, undiversified positions not because that is the smart allocation but because selling would set off a tax bill. That is capital frozen by the tax code rather than put to its best use. An accrual tax thaws it, and the money can move toward something more productive.

The deeper reason to take accrual taxation seriously is not really the revenue. It is the incentives. Right now the rational move for a very rich person is to never sell, ever, and to borrow against the position to fund a lifestyle without triggering a cent of tax. A founder holding $50 billion of one stock is not doing anything with that capital; the company already raised what it needed at the IPO. The code is quietly paying people to pile up concentrated positions and then sit on them, untaxed and inert.

An accrual tax, or even a credible threat of one, changes the math. Holding an appreciated asset now carries an annual cost. Some people would pay it and keep holding. Some would sell and put the money somewhere else, paying capital gains on the way out and re-entering the economy with a more diversified book. Some would hand over shares in kind, which flow into public hands. Some would give it away. Every one of those outcomes has the capital doing something. The one option that gets less attractive is the one we currently reward, parking a fortune in a single position forever and never paying tax on it.

This is a Pigouvian way of thinking about it, after Arthur Pigou, who argued for taxes that fix bad incentives rather than just raise money. The aim is not to make rich people poorer for its own sake. It is to make endless, idle accumulation in one concentrated position less appealing than the alternatives. That some of those alternatives, paying in kind, or selling and reinvesting, happen to throw off revenue or public ownership is a bonus. The core of it is behavioral more than fiscal.

And as with any Pigouvian tax, the real question is whether the fix is worth what it costs. Valuing private assets is hard, and the methods are contested. Avoidance schemes will keep evolving. The constitutional ground is unsettled, since Moore v. United States in 2024 dodged the question of whether income even has to be realized before it can be taxed. These are not small risks.

It is also worth saying what the evidence does not show. The Nordic countries tax capital about as heavily as anyone in the rich world, and they sit at or near the top of every serious innovation ranking. Denmark, Sweden, Finland, and the Netherlands match or beat the United States on the Global Innovation Index and its cousins. The idea that taxing capital must smother innovation just is not borne out, and the truth is messier and turns on what the money funds and how well the institutions run.

So, the honest summary. An accrual tax on concentrated wealth is a real tool aimed at a real distortion, and it could send revenue toward a Social Wealth Fund or anywhere else. Whether it pays off for society over the long run comes down to numbers economists still argue about, mostly the avoidance elasticity and the effect on investment. This piece maps the mechanism and shows how to build it carefully. It does not tell you to pull the trigger.

Strongest objections (Part IV). Valuing private assets. The split between tradable and non-tradable, plus the deferral charge, sidesteps most of the fight, because the illiquid assets never get marked at all. Capital flight and expatriation. The migration evidence shows far smaller effects than the intuition expects, the base is narrow and high-threshold with an exit-tax backstop, and US-held stock cannot physically move to dodge a US bill in the first place. Administrability. The base is a few thousand people, and it runs on brokerage reporting that already exists. Constitutionality. Moore v. United States in 2024 upheld the Mandatory Repatriation Tax and went out of its way not to decide whether realization is required, so the income-tax framing is the safest route, and the risk is on the table rather than buried. Lock-in. Accrual taxation actually loosens the realization system’s lock-in, which is its own source of misallocated capital.

What the data cannot see (Part IV). The behavioral elasticities are uncertain and the revenue range is wide. The stock of unrealized gains is estimated from distributional accounts rather than counted off tax returns. And the constitutional question is open. On whether an accrual tax is good for society on net, this piece takes no position. It lays out the mechanism, the size of the base, and the design choices that would make it defensible.

V · The Argument

Why is a public claim on the most powerful productive assets legitimate, and not just convenient? The policy chapters rest on a moral claim, and a claim like that should be argued out rather than smuggled in.

Luck versus desert

The gut feeling about a giant fortune is that it was earned, that the founder of a $1.77T company has it coming because he built the value. There is something real in that, but less than it first appears.

Take the philosophy first. Rawls talked about the “natural lottery,” the point that the things that let someone build an empire, the intelligence, the temperament, the country and decade and family they happened to be born into, were never earned in the first place. Behind a veil of ignorance, not knowing which life you would draw, you would not sign off on rules that let the lottery winners take almost everything. Dworkin sharpens this into a distinction between brute luck and option luck. We can let people keep what they win on gambles they chose to make, while treating the windfalls of sheer circumstance differently. Even Hayek, no friend of redistribution, admitted that the market pays people for being useful to others rather than for being morally deserving. The market is a signal, not a scoreboard for virtue.

Then the evidence. Robert Frank’s Success and Luck shows how much timing and chance matter out at the extreme tail, where a tiny early edge snowballs. The mobility research from Chetty and his colleagues shows how strongly your endpoint is predicted by your starting point, which is to say the self-made story leaves out a lot of inheritance and networks and timing. And the assumption that the hugely successful could just go do it again runs straight into regression to the mean and survivorship bias. For every founder who turned one improbable hit into an empire, there are many just as capable who hit the same wall and never got written about. None of this denies that skill and effort are real. It denies that skill and effort by themselves explain the tail, and so it denies that the tail’s rewards are some pure entitlement no public claim can touch, which matters all the more once you remember, from Parts I and II, that the public paid to build the runway.

”The rich are the job creators”

The strongest practical version of the objection is that this kills the goose. The rich are the job creators, the engine of investment, and you tax them at your peril. The demand-side answer is that jobs come from customers. A business hires when there is enough demand to justify it, and demand comes from people having money to spend. As for where the jobs actually come from, the data points at young firms rather than rich people. Haltiwanger, Jarmin, and Miranda find that it is a firm’s age, not its size, that drives net job creation, which means startups and young fast-growing companies, not incumbents and not the personal spending of the wealthy. A Social Wealth Fund that indexes broadly and pays a dividend does not starve that engine. It moves a slice of the returns to capital, returns the public often helped create, back to the public, and it leaves the price system and the reward for building a real company alone.

Economic democracy

The deepest argument is about power more than money. We take it for granted that political power has to answer to the people it governs. Yet most of us spend a third of our lives, half our waking hours, inside firms that Elizabeth Anderson calls private governments, hierarchies that direct our days in fine detail with almost none of the checks we demand of the state. Robert Dahl’s A Preface to Economic Democracy asks the obvious follow-up. If accountability is right for the government, why does it stop at the office door? Concentrated ownership of productive assets is concentrated power over the people who work there and the towns that depend on them.

A Social Wealth Fund does not democratize the firm on its own, but it is a start. Even non-voting public ownership creates a public claim and a public stake in how that power gets used, and it can be paired with real channels, worker or citizen seats on the board, of the sort Germany’s codetermination has run for decades. The research on codetermination is, tellingly, not a horror story. It finds roughly neutral-to-positive effects on productivity and stability, which cuts against the assumption that any loss of pure shareholder control spells ruin.

Strongest objections (Part V). Nozick’s entitlement theory, that if every step in building the fortune was fair, voluntary trade and no fraud, then the result is fair and redistribution tramples rights. The reply is that the steps were not purely private. Public R&D, procurement, spectrum, and rescue money were underneath them, so a public claim is not grabbing private product but recovering a public contribution. The Hayekian point about knowledge and incentives, that concentrated private control allocates capital better than any public body could, and that dulling the reward dulls the drive to build. This is the serious one, and it is why the design keeps the price system, keeps the founder’s upside large, uses non-voting stakes, and indexes instead of directing. Public-choice skepticism, that government will capture the fund and waste it. Answered the same way as in Part III, with Norway-style arm’s-length governance, a hard legal lock, and a visible dividend that gives every citizen a reason to guard it.

What remains contested. Clear away the empirical fights and a real disagreement about values is left standing, over how much weight to give a person’s claim to what they earned against everyone’s claim on a product that was partly collective. The data can settle the job-creator and capital-flight questions. It cannot settle whether the tail of fortune is deserved. My read is that it mostly is not, but that is a moral judgment, reasonable people land elsewhere, and the design is built to hold up even for someone who weighs desert more heavily than I do.

Appendix · Methods, data, and reproducibility

Single source of truth. Every headline figure lives once, in data/facts/ipo_facts.yaml, and is read by both this prose (via an injected spacex_facts.json) and the dashboards. Prose and dashboards therefore cannot disagree on a number. Every figure traces to a primary source in the repository’s SOURCES.md.

Pipeline. Six numbered stages, fetch → clean → model → score → dashboards → facts, run end-to-end in a few seconds. Stage 00 pulls from keyless sources (Yahoo Finance for SPCX and 30-plus peers and benchmarks, FRED for CPI, USAspending.gov for SpaceX’s federal awards), validates each pull, and falls back to the last good snapshot on failure so a bad fetch never overwrites good data. The whole thing re-runs weekly via GitHub Actions and commits any changes, which triggers the site’s existing deploy.

Methods by part. Part I covers valuation by sum-of-the-parts, a reverse-DCF that solves for the implied revenue CAGR, and an explicit scenario DCF, plus the first-day-pop percentile against a curated mega-IPO set, market-model abnormal returns (estimation window then cumulated CARs) for the peer event study, a lockup and index-flow supply-demand model, and a likelihood-times-impact risk register. Part II is USAspending aggregation plus an illustrative TARP “kept-it” counterfactual, where realized proceeds are reinvested in a market index from disposition to today, which is descriptive rather than a causal estimate. Part III is a deterministic fund-growth simulator calibrated to GPFG’s long-run real return and spending rule. Part IV is an accrual-tax revenue model with avoidance and loss-refund parameters, plus the buy-borrow-die and symmetry illustrations.

Honesty rails. Every part carries a Strongest objections response and a What the data cannot see note. Curated or illustrative inputs (the comparable set, the TARP figures, the wealth-distribution buckets) are labelled as such. Analyst assumptions (multiples, scores, elasticities) are exposed as dashboard sliders rather than hidden.

The living tracker. Four falsifiable Part I predictions were frozen at publication into predictions.yaml. The weekly run scores them against the realized tape and appends to a public changelog, so the calls are graded rather than quietly forgotten.