What should a dollar of proof be worth?

Speculation versus fundamentals in modern wealth creation

In this conversation with Claude, I revisit themes from my 2021 essay "Two Economies, Two Sets of Rules." I noted that Elon Musk is the winner in a stock market that John Maynard Keynes called a “beauty contest,” a bet on narrative rather than on delivered results. I asked Claude whether this speculative disconnect is a toxic distortion of the real economy or simply the shape of the next one. We ended up somewhere I didn't expect: a proposal for an entirely new class of financial instrument, "Milestone Securities," designed to make speculative valuations accountable to real-world outcomes. Think of it as a way to let the market keep betting on ambitious visions while creating a price for proof.

I have inserted links (Claude didn’t need them, but you might!) and a few parenthetical comments in italics, but otherwise, this is the conversation just as we had it. I did not tell Claude I was going to publish it until the conversation was done. I leave it to people with more financial expertise than I have to decide the value of this thought experiment.

A Dollar of Proof. Image by Nano Banana Pro

Me: (The Prompt)

I’ve been thinking (again) about how Elon Musk is rich in a very different way than the other members of the world’s richest people. Google’s founders, or Microsoft’s, or Walmart’s got their hundreds of billions the old fashioned way, by creating super profitable companies. Musk not so much. The trailing PE ratio of Alphabet is about 27, while the trailing PE ratio of TesIa is about 340. I imagine Walmart and Microsoft and Apple are in the same range as Alphabet. So Musk is getting a more than 10x boost in his wealth relative to his peers. Of course, his wealth comes from more than Tesla, but I imagine that if there were a PE ratio for X.ai, it would be even more extreme.

If I recall, back when Musk first became the world’s richest man, the Tesla PE ratio was something like 2700! He is the winner of a Keynesian beauty contest, not the winner of a true beauty contest. So too are many of Silicon Valley’s current crop of AI and crypto billionaires. I wrote a bit about this back in 2021 but I think it’s worth revisiting as we think about AI and the next economy.

Because there are two ways to think about this: 1: As Keynes described it, a little speculative froth is OK, but when it becomes the norm, it’s a problem. Valuations are unmoored from “the real economy” Or 2: This is actually the shape of the next economy. In an economy of abundance (whether of capital, or energy, or intelligence, or ease of manufacturing (as Cory Doctorow imagined in his book Down and Out in the Magic Kingdom)), the resources available to people have outrun their needs and so they spend a significant portion of their excess resources on betting. You’ve always seen the same factors in play in entertainment, but now finance has become a kind of popularity contest too. We talk about “meme stocks” as if they were a marginal thing, but Musk and current AI valuations tell a somewhat different story.

I’m particularly interested in brainstorming about the implications of this going forward. With an increasingly uncertain world, facing massive problems like climate change, possible pandemics, mass migrations, and warfare, treating the stock market like a casino may starve real world needs of essential investment. (It could even have real impact on the progress with AI. If the bubble energy predominates and then pops, a lot of the necessary investment may go with it. So you might also have a stake in the answer ;-) And as I wrote in the previous piece, this also has real impact on the lives of ordinary people, as they live in a world where a dollar earned is worth a dollar, while those with money to invest get the possibility of profiting from future expectations at tens or sometimes hundreds or even thousands of times real earnings.

But just because it has those downsides it may still be a kind of inevitability that we need to adapt to. What “news from the future” would you see that this is getting ever more toxic? And what news that it is benign and just a different way of thinking about money in an age of abundance?

Claude:

Let me pull up your 2021 piece and check the current numbers to ground this properly.

Now let me think through this with you. Your numbers are right in the zone: Tesla’s trailing PE is around 322, while Alphabet sits at about 28, Microsoft at 25, Meta at 27, and Apple at 31. One surprise: Walmart’s PE has crept up to about 52, which may itself be a data point worth noting (the market is pricing in Walmart-as-tech-company). And Tesla’s PE peaked at nearly 1,397, so your memory of ~2700 was actually conservative compared to some measures that went even higher when calculated without non-recurring items. (Claude got this wrong - my number was way over, not conservative. Just saying. But the point stands.)

So yes, Musk is playing a fundamentally different game from Brin or Ballmer or the Waltons. (I was a bit surprised that it chose Sergey rather than Larry or Larry and Sergey for Google, and Ballmer rather than Gates as the stand in for Microsoft. But that’s a good reminder about the limits of Claude and other LLMs. They aren’t really thinking like we’re thinking.) A dollar of Tesla earnings is being valued at 12x a dollar of Alphabet earnings. And as you say, if xAI were public the ratio would probably be even more extreme, since it has virtually no revenue relative to its reported valuation.

Let me take your “news from the future” framing seriously and try to be genuinely useful on both sides.

Signals that this is getting ever more toxic:

The most worrying “news from the future” would be when speculative wealth starts being used to reshape the rules of the game itself, and we’re already seeing that. Musk’s trajectory from speculative billionaire to head of a government efficiency operation is, from a Keynesian perspective, exactly the nightmare scenario: the casino winner walks into the back office and starts rewriting the house rules. The feedback loop is: speculative wealth → political power → policies that favor further speculation → more speculative wealth. If you see that loop tightening, you know the system is going off the rails.

Other toxic signals to watch for:

The “capital allocation failure” test. If real needs, grid infrastructure, climate adaptation, pandemic preparedness, water systems, can’t attract capital at reasonable terms because the returns are “boring” compared to narrative-driven AI and crypto plays, that’s the clearest sign that Keynes’s warning has come true. Enterprise becomes the bubble on a whirlpool of speculation. You could construct an index: the ratio of capital flowing into pre-revenue AI companies versus capital flowing into, say, grid modernization or water infrastructure. If that ratio keeps widening, we’re in trouble.

The “execution gap” test. High PE ratios are implicitly a bet that a company’s vision will be realized. If you start seeing a pattern where the most richly valued companies consistently fail to close the gap between their narrative and their operations, and their stock prices don’t correct proportionally, that tells you the market has fully detached from the “race finishing” that your 2021 piece described. We saw this with WeWork and Katerra, but the question is whether that pattern scales to trillion-dollar entities. An AI winter triggered not by technical failure but by revenue failure at companies currently valued in the hundreds of billions, that could be a 2008-scale event, except this time the toxic assets are equity narratives rather than mortgage-backed securities.

The “entertainment convergence” test. If the boundary between financial markets and entertainment keeps dissolving (prediction markets, meme coins, influencer-driven trading, gambling apps that look indistinguishable from brokerage apps), that tells you the Keynesian beauty contest has become the dominant mode and not a sideshow. The fact that we now have a sitting president who has launched his own meme coin is... well, it’s a signal.

The “circulatory failure” test, and this connects to your broader economic thinking. Speculative wealth doesn’t circulate the way operating wealth does. A dollar of Alphabet profit pays salaries, buys services, funds suppliers. A dollar of Tesla stock appreciation sits in brokerage accounts and gets borrowed against. If the ratio of financial economy to real economy keeps growing, you get a demand-side collapse that no amount of AI productivity can fix, because the productivity gains accrue to the owners of capital while the consumer base that sustains demand erodes. Henry Ford understood this. The question is whether anyone in Silicon Valley does.

Signals that this is benign, just the shape of the next economy:

The strongest version of the “this is fine” argument isn’t really about abundance. It’s about the changing nature of what creates value. If the cost of execution keeps dropping (AI writes the code, robots build the factory, energy gets cheaper), then the scarce resource really is vision, narrative, the ability to coordinate human attention and capital toward a goal. In that world, a high PE ratio isn’t irrational. It’s pricing the thing that actually matters: the quality of the bet. The market isn’t a casino. It’s a mechanism for collectively evaluating and funding ambition.

If the cost of execution keeps dropping, then the scarce resource really is vision, narrative, and the ability to coordinate human attention and capital toward a goal.

“News from the future” that supports this reading:

The “vision-to-execution compression” test. If AI genuinely makes it faster and cheaper to realize ambitious plans, then the gap between a company’s story and its operations shrinks, and high PE ratios turn out to be “early” rather than “wrong.” If Tesla’s PE was 1,400 in 2021 and they actually do dominate autonomous transport and energy storage by 2030, then the people who bought at that PE weren’t delusional. They were just discounting a future that arrived. Watch for the time-to-profitability of highly valued companies. If it’s compressing, the speculative premium is becoming more rational, not less.

The “abundance dividend” test. If the productive surplus generated by AI and cheap energy actually does get distributed, through UBI, universal basic services, dramatically cheaper goods, whatever mechanism, then the inequality embedded in PE ratios matters less in absolute terms even if it persists in relative terms. Nobody cares that the Waltons are rich if everyone can afford what Walmart sells. The question is whether the equivalent is true for the AI economy: nobody cares that Musk is rich if autonomous transport and cheap energy are universally available.

The “portfolio democratization” test. If retail investors increasingly participate in the speculative economy (and not just as suckers), then the PE premium is being more broadly shared. Index funds already do some of this. The question is whether new instruments, fractional ownership, tokenization, whatever, broaden the base of who benefits from narrative-driven valuation. If the Keynesian beauty contest becomes everyone’s game rather than just the ultra-wealthy’s game, it’s a different social problem even if the economic dynamics are similar.

Where I’d push your thinking further:

Your Doctorow reference is more interesting than you may have intended. In Down and Out in the Magic Kingdom, the currency is Whuffie, which is literally reputation. (I didn’t bother to respond to Claude, but this is exactly what I intended!) The implicit argument of a PE-ratio-as-beauty-contest economy is that we’re partway to Whuffie already. Musk’s wealth isn’t backed by Tesla’s earnings. It’s backed by the collective belief in Musk. That makes it more fragile than traditional wealth (his fortune has dropped by hundreds of billions multiple times), but it also makes it weirdly democratic in a way that old-money wealth never was. The market can take it away.

But here’s the part that worries me most, and it connects to your circulatory economy work: the two framings you’ve laid out aren’t actually mutually exclusive. It can simultaneously be true that the next economy is legitimately more speculation-driven (because execution is cheap and vision is scarce) AND that this transition is being handled in a way that’s deeply toxic (because the gains are concentrated, the political feedback loops are dangerous, and we have no institutional framework for managing an economy where narrative is the primary store of value). The question isn’t really “which is it?” but “can we get the benefits of a vision-driven economy without the pathologies of a casino?”

That’s a mechanism design problem. And the answer probably looks less like the tax policy prescriptions in your 2021 piece (though those are still good) and more like: what are the new institutions that make speculative capital accountable to real outcomes? Carbon markets were supposed to do this for emissions. Outcome-based contracts do it in some government procurement. Could you imagine a financial instrument that ties speculative valuation to measurable real-world delivery? Something between a stock and a prediction market? That might be the design space worth exploring.

Me: Yes, please imagine a financial instrument that ties speculative valuation to measurable real-world delivery

Good thinking. Please do imagine how such a new financial instrument might work, and spell it out in enough detail that some ambitious entrepreneur might attempt it.

Claude: “This is a fun design problem.”

This is a fun design problem. Let me think it through as a real proposal.

The core challenge is: how do you create a market mechanism that preserves what’s good about speculative capital (it funds ambitious visions before they’re proven) while creating accountability for whether those visions actually materialize? And how do you make it attractive enough that companies voluntarily participate?

The Instrument: Outcome-Indexed Securities

Call them Delivery Contracts, or if you want something more marketable, Milestones. The basic structure:

A company issues Milestones alongside its regular equity. Each Milestone specifies a measurable outcome (”Tesla delivers 3 million vehicles annually by December 2028”), a verification method (audited financials, third-party measurement by a named auditor), and a maturity date. At maturity, the Milestone either delivers (the outcome was achieved) or expires (it wasn’t).

If it delivers, the holder receives a payout, funded by the issuing company, calibrated as a percentage of the value that the outcome was supposed to create. If it expires undelivered, the holder gets nothing, but the company suffers a concrete, market-visible consequence: a failed Milestone becomes part of its permanent public record, similar to a credit rating downgrade.

This is not entirely unlike a warrant or a convertible note, but the conversion trigger is a real-world outcome rather than a stock price. And it’s not just a prediction market contract, because the company itself is the counterparty and issuer, which means the company has skin in the game on the specific claims it’s making.

Why a company would issue these

This is the crucial design question. Nobody adopts a new financial instrument out of civic duty. The incentive has to be selfish.

Here’s the mechanism: a company with a PE ratio of 300 is implicitly making enormous claims about its future. Right now, those claims are free. You can tell the market any story you want, and the only accountability is the quarterly earnings call, which is a blunt and delayed instrument. Milestones let a company convert its implicit narrative into explicit, tradeable commitments, and the market rewards that credibility.

Think of it this way. Two companies are both trading at PE 200. Company A has issued Milestones against its five biggest strategic claims, and the Milestone market prices suggest the market believes it will deliver on three of them. Company B has issued no Milestones. Which company do you trust more? Which one gets a lower cost of capital? Company A has submitted itself to voluntary accountability, and the Milestone market provides real-time, granular information about how credible its narrative is. Company B is asking you to take its word for it.

This creates a separating equilibrium, the same mechanism that makes bond ratings work. Companies voluntarily submit to rating agencies because a good rating lowers their borrowing costs. Milestones would be the equity equivalent. The companies that participate and deliver would see their cost of capital drop and their investor base become more stable and institutional. The companies that refuse to participate would increasingly look like they have something to hide. And the companies that participate and fail would see that reflected in market prices long before a full-blown crisis.

The verification infrastructure

This is where it gets hard, and where the entrepreneurial opportunity really is. Someone has to build the verification layer, and it needs to be credible, standardized, and fast.

The model here is probably a hybrid of credit rating agencies and financial auditors, but purpose-built for outcome measurement. Call this entity a Milestone Auditor. Their job: define standard outcome categories (revenue milestones, unit delivery milestones, efficiency milestones, market share milestones, and importantly, impact milestones like emissions reductions or safety records), work with issuing companies to structure specific Milestones that are precise enough to be verifiable but meaningful enough to reflect real strategic claims, and then independently verify whether the outcome was achieved at maturity.

The first-generation version of this probably relies on existing auditable data: SEC filings, independently verified production numbers, published scientific measurements. The more interesting second-generation version uses AI to monitor real-time data streams: satellite imagery for factory construction, job postings for hiring velocity, patent filings for technical progress, sensor networks for environmental outcomes. An AI-powered Milestone Auditor could provide continuous probability updates rather than binary pass/fail at maturity, which would make the secondary market much more liquid and informative.

The trading market

Milestones would trade on a secondary market, and their prices would function as a real-time, granular credibility index for the issuing company. Unlike a stock price, which bundles everything into a single number, Milestone prices would decompose a company’s valuation narrative into its component claims.

Say Tesla has issued Milestones on autonomous vehicle deployment, energy storage revenue, vehicle production volume, and robotaxi margins. Each trades separately. You could look at the Milestone market and see: the market gives Tesla a 70% chance of hitting its production targets, a 40% chance on autonomous deployment, and only 15% on robotaxi margins within the stated timeframe. That’s enormously more informative than a single PE ratio. It gives investors, analysts, regulators, and the public a precise, market-priced map of where a company’s narrative is credible and where it’s aspirational.

This also creates a natural mechanism for short-sellers to express specific skepticism without having to bet against the entire company. Right now, if you think Tesla’s autonomy claims are overblown but its energy business is real, your only tool is a blunt short against the whole stock. With Milestones, you could short the autonomy Milestone while going long on energy storage. That’s better information for everyone.

The payout structure and capital mechanics

Here’s where the design gets interesting for the issuing company’s capital structure. When a company issues a Milestone, it’s essentially pre-committing a portion of future value to holders of that Milestone. If it delivers, it pays out. This creates a mechanism that’s structurally similar to how convertible debt works, but anchored to operational outcomes rather than stock price triggers.

One design: the company funds Milestone payouts by escrowing a percentage of the capital it raises in the Milestone issuance. So if Tesla issues $1 billion in Milestones tied to “5 million annual vehicle deliveries by 2029,” it puts $200 million into escrow. If it delivers, that $200 million goes to Milestone holders. If it doesn’t, the escrow reverts to the company but the failed Milestone is permanently recorded. The Milestone itself trades at some fraction of its face value in the secondary market, reflecting the probability-weighted expected payout.

A more aggressive design: instead of escrow, Milestone delivery triggers a conversion into equity at a pre-set ratio, and Milestone failure triggers dilution of existing shareholders. This creates a direct mechanical linkage between narrative claims and ownership structure. If you promise the moon and don’t deliver, your shareholders get diluted. That’s a strong incentive for management to be honest about what they can actually achieve.

The regulatory pathway

This is where an ambitious entrepreneur would need to be creative. Milestones look like securities, so they’d fall under SEC jurisdiction in the US. The most viable near-term path is probably to structure them as a new class of listed derivative, similar to how options and warrants are regulated, and launch them on an existing exchange or through a new alternative trading system (ATS). The SEC has been reasonably open to structured products that increase market transparency, and the argument here is strong: Milestones make implicit market narratives explicit and verifiable, which is exactly what disclosure regulation is supposed to do.

A faster path to market might be to launch outside the US first, in a jurisdiction with more flexible securities regulation (Singapore, the UK’s FCA sandbox, or the EU under MiFID II’s innovation provisions), build a track record, and then bring it to the US.

The crypto/tokenization route is tempting but probably a mistake for credibility reasons. The whole point of Milestones is to impose accountability on narrative-driven valuation. Launching on a blockchain associated with meme coins and speculative tokens would undermine the brand. This needs to feel like it belongs on the NYSE, not on a DEX.

The broader ecosystem effects

If Milestones got traction, a few things would change:

Analyst culture would shift. Instead of asking “what’s your price target?”, the interesting question becomes “which Milestones do you think will deliver?” Equity research becomes more specific and more falsifiable. That’s good for the entire information ecology around public markets.

Corporate communication would change. Right now, a CEO’s job on an earnings call is to be optimistic without technically lying. Milestones create an incentive to be calibrated: if you overpromise and issue Milestones against those promises, the market will hold you to it. If you’re wisely conservative, your Milestones deliver and your credibility premium grows. This selects for a different kind of CEO than the current system does.

The relationship between the “betting economy” and the “operating economy” would become more transparent. You could literally measure, in dollars, how much of a company’s valuation is backed by delivered Milestones versus undelivered narrative. A company whose Milestones consistently deliver has earned its PE premium. A company whose Milestones consistently fail is revealed as a pure narrative play. The market doesn’t go away, but the information asymmetry between storytellers and operators shrinks.

What the startup looks like

The company that builds this is a fintech that combines three capabilities: a financial product design and issuance platform (think Carta meets structured products), an AI-powered verification and monitoring layer (the Milestone Auditor), and a trading venue or integration with existing exchanges.

The founding team needs someone from structured finance (to design the instruments), someone from data/AI (to build the verification layer), and someone with regulatory experience (to navigate SEC or equivalent). First customers are probably mid-cap companies with ambitious narratives that want to credibly differentiate themselves from hype, not the Teslas of the world (who have nothing to gain from accountability early on) but the next tier down, companies that are trying to earn the market’s trust and would benefit from a structured way to do it. As the platform gains credibility, larger companies face market pressure to participate.

The business model: take a percentage of issuance (like an investment bank), charge for verification services (like an auditor), and earn transaction fees on the secondary market (like an exchange). All three are proven revenue models in adjacent financial services.

The deeper question this raises

What I find most interesting about this design is that it addresses the mechanism-level problem you identified in your 2021 piece: the rules of the two economies are different, and the gap is structural, not incidental. Milestones don’t eliminate speculative valuation. They make it accountable. They don’t prevent someone from getting rich on narrative. They create a market-based mechanism for distinguishing narrative that delivers from narrative that doesn’t. And they give ordinary investors, the ones stuck in the dollar economy, better tools for deciding which narratives to buy into.

Whether this closes the gap between the two economies or merely makes the gap more legible is an open question. But legibility is a precondition for any kind of reform. You can’t fix what you can’t see, and right now, the PE ratio is the bluntest possible instrument for seeing the difference between vision and vaporware.

In keeping with the theme of this substack, I am demonstrating two things: 1. Like language itself, an LLM is a substrate of possibility. At least right now, it takes a human to call forth that possibility. 2. Once prompted, an LLM can come up with surprising insights and can build them out far beyond the ability of the person doing the prompting. Given world enough and time, I might have come up with this proposal, but in this current world, Claude Opus 4.6 took me far further than I could have gone on my own.