Does a transformative technology necessarily justify today’s valuations?

The key questions are never simply whether the technology will work. The real questions are:

· Who will capture the value?

· When will that value be realized?

· How much of it has already been priced into financial markets?

These are the questions investors should be asking about AI today.

Physical AI May Not Matter Much in the Short Run

When people discuss AI, they are usually referring to large language models (LLMs). That is understandable. LLMs have achieved far more visible success than most other branches of AI, particularly physical AI.

One reason is straightforward: language data is abundant.

Over the past several decades, humanity has generated an enormous amount of text on the internet, much of it freely available for training. We also possess vast collections of images and videos, which help explain the rapid progress in image and video generation.

Physical AI faces a very different challenge.

Consider robotaxis and humanoid robots. Both require enormous amounts of training data. Robotaxis already benefit from substantial driving datasets, yet the industry continues to struggle with the final step toward widespread deployment after years of hard work.

Humanoid robots face an even greater obstacle: we simply do not possess comparable quantities of high-quality training data.

This may partially explain why technology companies are so interested in human-shaped robots. In principle, human-like machines can learn from human-generated data. If machines are built differently, generating training data may become far more expensive. Even so, data scarcity remains a major bottleneck.

Robotaxis also face a simpler physical environment. They primarily solve a two-dimensional navigation problem on roads, whereas humanoid robots must operate in fully three-dimensional environments involving manipulation, balance, and complex interactions with objects.

Moreover, many of the occupations frequently discussed in connection with humanoid robots are not obviously large labor markets. Manufacturing and construction employ far fewer workers than many service industries in the United States.

The larger economic opportunities may lie in healthcare, elder care, and household services. Yet these applications require extremely high levels of reliability and safety. Before we can assess their economic potential, we must first solve the underlying data problem.

For this reason, I am considerably more optimistic about LLMs than physical AI over the near term. When discussing current AI valuations, I place relatively little weight on physical AI.

From this point forward, I use “AI” to refer primarily to LLMs.

The Critical Question: Who Captures the Value?

Suppose AI succeeds. The next question is whether the developers of frontier models will capture most of the resulting economic value.

I am not convinced.

Before the AI boom, the Magnificent Seven largely occupied businesses protected by powerful competitive advantages. These advantages arose from network effects, platform ecosystems, switching costs, and economies of scale.

For example, Facebook illustrates the power of network effects. The platform became more valuable as more users joined because each new user increased the value of the network for everyone else. This self-reinforcing dynamic made it difficult for competitors to attract users and helped Facebook establish a dominant market position. Apple, meanwhile, benefits from a tightly integrated ecosystem that makes switching costly for many consumers. These structural advantages support long-run pricing power.

The economics of LLMs look much less certain.

First, entry barriers may be lower than many investors assume. DeepSeek demonstrated that a relatively resource-constrained organization could build a model approaching the performance of leading frontier systems.

Second, network effects appear weak. Users frequently switch between models. Many individuals and firms use multiple AI systems simultaneously. The market remains fluid, and it is far from obvious that any provider will achieve durable monopoly power.

Third, open-source competition is becoming increasingly important. Open-source models are often cheaper and may be preferred by enterprises concerned about privacy, security, and control over proprietary data.

Fourth, AI development is extraordinarily talent-intensive. Even if frontier-model developers eventually earn substantial profits, a meaningful fraction of those profits may flow to researchers and engineers rather than shareholders.

For these reasons, the LLM industry currently looks a lot more competitive than monopolistic.

If that assessment is correct, much of AI’s economic value may accrue elsewhere.

The most obvious beneficiaries are the suppliers of critical infrastructure: semiconductor designers, chip manufacturers, memory producers, and equipment suppliers. Companies such as NVIDIA, AMD, TSMC, Samsung, SK Hynix, and ASML may possess more defensible competitive positions than many application-layer AI firms.

But this raises another question.

If LLM developers ultimately struggle to earn attractive profits, can current levels of spending on data centers and AI infrastructure be sustained? If not, what are the implications for the broader AI ecosystem?

One might argue that even if today’s frontier-model developers fail, newly successful AI companies could simply inherit the existing data-center infrastructure. However, this view overlooks the industry’s circular financing structure. Major technology and infrastructure companies are not merely suppliers; they are also major investors in AI startups, which in turn spend much of that capital on the investors’ own chips and cloud services. If today’s frontier-model developers run into trouble, their suppliers and capital providers are likely to suffer as well. The fortunes of the two groups are therefore more tightly linked than they may initially appear.

The Timing Problem

Even if AI ultimately creates enormous value, investors must still ask when that value will appear.

History offers a cautionary lesson.

Electricity took decades to transform productivity. Computers took decades to transform productivity. In 1987, Robert Solow famously observed:

“You can see the computer age everywhere but in the productivity statistics.”

The same possibility exists for AI.

Recent studies document some time savings from AI adoption. Yet measurable productivity gains remain limited. Firms are still experimenting with deployment strategies, organizational redesign, and new workflows.

In other words, companies increasingly recognize the usefulness of AI. What they have not yet figured out is how to reorganize themselves to fully exploit its potential.

The benefits of AI may eventually be enormous—but they may arrive more slowly than current market expectations imply.

Is There a FOMO Component to AI Demand?

You can judge whether the following dynamic is important.

Firms adopt AI because they fear being left behind. Frontier-model developers invest aggressively because they believe AI may become a winner-take-all industry and that they can sustain a technological lead. Investors, meanwhile, reward AI spending because they fear missing the next technological revolution. The result is exceptionally strong demand for AI infrastructure, allowing bottleneck suppliers such as Nvidia and Micron to earn extraordinarily high gross margins—roughly 75% and 85%, respectively. Such margins imply substantial pricing power: an 85% gross margin means that a product costing $1 to manufacture sells for about $6.67.

I do not think such a high gross margin will sustain, given that memory manufacturing generally has lower technological differentiation than leading-edge logic chips, making sustained monopoly-like margins harder to justify over the long run.

Moreover, the industry is no longer limited to the three established leaders—SK Hynix, Samsung, and Micron. Chinese memory manufacturers, including ChangXin Memory Technologies (CXMT) and Yangtze Memory Technologies Corp. (YMTC), have expanded rapidly and are reportedly preparing for Hong Kong IPOs. Their emergence suggests that competition is likely to intensify rather than diminish over time.

It is therefore difficult to know how much of today’s surge in AI-related spending reflects durable demand and how much reflects experimentation/competition driven by fear of missing out.

That distinction matters because experimentation and intense competition are temporary. Sustainable demand is not.

The China Factor

One aspect of the AI discussion that receives surprisingly little attention is competition from China.

China now hosts a large and rapidly growing ecosystem of AI models, many of which are inexpensive and open source.

Even if geopolitical tensions eventually create some degree of segmentation between the U.S. and Chinese AI ecosystems, investors should not ignore the rest of the world.

Outside the two largest economies, American and Chinese models will increasingly compete for users, developers, and enterprises.

Open-source models (whether Chinese or not) may also continue to gain market share within the United States.

The result could be a more competitive global landscape than many current valuation narratives assume.

AI as a Systemic Risk

Another issue that receives insufficient attention is systemic risk.

AI-related firms now account for an extraordinary share of total equity-market capitalization. They have also become major consumers of capital and major borrowers.

As a result, AI has evolved from a sector-specific story into a market-wide risk factor.

During the internet bubble, technology represented a much smaller share of household wealth and financial markets than today’s AI ecosystem does.

The concentration of market value in a handful of AI-linked firms means that future disappointments could have broader macroeconomic and financial consequences.

There are at least two important implications. First, at some point, regulators may begin viewing AI concentration not merely as a technology issue but as a financial stability issue, much as they do systemically important financial institutions (the too-big-to-fail problem).

Second, many AI-related businesses are closely tied to the broader economy and financial markets. In asset-pricing terms, they likely have high betas and therefore warrant relatively high discount rates. Higher discount rates, all else equal, reduce the present value of future cash flows.

Technology vs. Investment

None of these arguments implies that AI is overhyped or destined to disappoint. They simply suggest that investors should distinguish between technological success and investment success. History repeatedly shows that transformative technologies do not automatically generate extraordinary shareholder returns for their earliest leaders. Competition, timing, and valuation matter just as much as innovation itself. Technology always progresses, but the path of valuations is rarely smooth.

Bottom Line

AI may ultimately justify today’s optimism.

Or it may not.

The future of AI depends not only on technological progress but also on competition, profitability, adoption, timing, and market expectations.

Don’t forget that AI is still a big bet.

That is what investors should keep in mind.

Since the beginning of Donald Trump’s second term, the President has repeatedly pressured Federal Reserve Chair Jerome Powell to lower interest rates. “Mr. Too Late” became one of Trump’s favorite nicknames for Powell, reflecting his view that the Fed was moving too slowly to cut rates.

The pressure extended beyond public criticism. In November 2025, federal prosecutors in Washington, D.C., opened a criminal investigation into Powell concerning the renovation of the Federal Reserve’s headquarters and whether he had made false or misleading statements to Congress about the project’s scope and cost. Many observers viewed the investigation as part of a broader effort to pressure the Fed into adopting a more accommodative monetary policy.

Three Responses

1. Monetary policy remained largely unaffected.

Despite intense political pressure, the Federal Open Market Committee (FOMC) continued to make decisions based on its assessment of economic conditions rather than presidential preferences.

2. Powell chose to stay.

Powell himself responded in an unusual way. Rather than stepping away after his term as Chair, he announced that he would remain on the FOMC as a regular governor. This broke a long-standing convention under which former Fed Chairs typically resign their FOMC membership after leaving the chairmanship. The decision was widely interpreted as a signal that he would continue defending the institution’s independence.

3. Public support for independence strengthened.

On June 1, 2026, Powell received the John F. Kennedy Profile in Courage Award at the John F. Kennedy Presidential Library in Boston. He was honored for his defense of central bank independence in the face of sustained political pressure.

Perhaps most strikingly, Powell’s successor did not appear eager to accommodate the White House. At his first FOMC meeting, new Fed Chair Kevin Warsh emphasized both Fed independence and price stability. For anyone who follows monetary policy closely, the emphasis on “price stability” carries a distinctly hawkish tone—it signals a willingness to take inflation risks seriously. This surprised the market: the S&P 500 fell by more than 1% following the meeting.

And it is worth remembering one final fact: both Powell and Warsh were nominated by Trump.

In my view, the past few years have amounted to a real-world stress test of the Federal Reserve's independence. The result is that the institutional arrangements protecting that independence have proven remarkably resilient. Far from becoming less confident in the Fed’s autonomy, I have become more confident in it.

I struggle to think of many other major central banks that enjoy a comparable degree of independence. The European Central Bank may be one exception, largely because there is no central European government or single European government capable of exerting the same degree of political pressure.

The broader lesson is that institutional independence is not demonstrated when political leaders agree with policy decisions. Any central bank can appear independent when there is no conflict. True independence is revealed when political leaders strongly disagree with policy, exert substantial pressure, and yet the institution continues to act according to its mandate.

By that standard, the Federal Reserve has just passed one of the most demanding tests in its history.

Nothing is perfect, but what we have is pretty good.

Suppose we lived in a world with a single global government—or, in economists’ terminology, a social planner—and a population of 1.5 billion people. (I’ll explain that number shortly.)

Would you support the following proposal?

The government taxes each person $33.33 per year and uses the proceeds to develop electric vehicles, reusable rockets, and satellite internet.

Why 1.5 billion people? Roughly 1 billion people live in developed economies, and I assume another 500 million people in developing countries could reasonably afford such a contribution. Don’t ask me how I arrived at this number—it is more intuition than science, and the exact number is not important.

Why $33.33 per year? Using a 5% discount rate (roughly in line with the long-term U.S. Treasury yield), a perpetual payment of $33.33 per year has a present value of about $666.67. Multiply that by 1.5 billion people, and you get approximately $1 trillion.

If you don’t like the idea of paying an additional tax, imagine it differently. The government could simply redirect $33.33 from the taxes you already pay. In practice, we would probably ask wealthier individuals to contribute more, while many lower-income individuals might pay little or nothing at all.

By now, you probably know where I am going with this.

This thought experiment is useful when thinking about the debate over whether Elon Musk is too rich. The real-world math is obviously different. In reality, Elon Musk did not invest $1 trillion, nor were Tesla and SpaceX primarily funded by taxpayers. Instead, investors voluntarily supplied the capital—and most of them have been richly rewarded for doing so.

Why I’m Thinking About This Now and A Simple Question

Over the past few days, it seems everyone I know has been talking about SpaceX and its IPO. I have been paying close attention myself and have already written about SpaceX twice—here and here.

There is no doubt that Elon Musk is one of the most controversial figures in the world. Some people admire him; others strongly dislike him. Among my friends and students, the divide is remarkably clear.

I would suggest taking a step back and setting aside political ideologies for a moment. Ask yourself a simple question: Has Elon Musk contributed to society? Has he made our lives better?

My answer is a clear yes. But that is not what fascinates me most.

The Real Miracle

What fascinates me even more is that our economic system is capable of funding extraordinarily ambitious ideas—ideas that often appear unrealistic, if not impossible, at the outset. As a society, we have found a way to channel capital toward people like Elon Musk and to reward them when they succeed.

Whether it is electric vehicles, reusable rockets, satellite internet, or future projects we cannot yet imagine, the broader lesson is not just about Elon. It is about the remarkable ability of markets and institutions to mobilize resources in pursuit of transformative innovation.

That, to me, is the real miracle.

He provided few details beyond describing it as a great deal.

Before investors and policymakers get too excited, however, there is a more fundamental question:

Why would either side trust the other to honor the deal?

That question may sound mundane. In reality, it sits at the heart of nearly every failed U.S.-Iran negotiation over the past four decades.

Read more

· At SpaceX’s $135 offering price, Elon Musk’s SpaceX stake is worth approximately $819 billion.

· Musk’s Tesla holdings, including both shares and vested options, were worth approximately $297 billion on June 3.

· Counting only SpaceX and Tesla, Musk’s wealth reaches roughly $1.12 trillion.

· Using the secondary-market price of SpaceX shares instead of the offering price pushes his net worth even higher.

· Regardless of the valuation method, Musk appears poised to become the first trillionaire in human history.

SpaceX: The Largest Piece

Based on SpaceX’s prospectus, the company’s Founder, Chief Executive Officer, Chief Technical Officer, and Chairman of the Board will own:

· 849,494,440 shares of Class A common stock

· 5,219,053,075 shares of Class B common stock

immediately following the offering.

At the offering price of $135 per share, those holdings are worth approximately $819.3 billion.

SpaceX is now the single largest contributor to Elon Musk’s wealth.

Tesla: Another $300 Billion

Musk’s second-largest asset is his stake in Tesla.

The most recent information comes from an amended Schedule 13G/A filed on April 23, 2026. According to that filing, Musk, through the Elon Musk Revocable Trust, owns 413,152,109 Tesla shares. He also holds options on 303,960,630 shares, almost entirely the remaining portion of his famous 2018 CEO Performance Award.

Those options have an exercise price of just $23.34 per share.

On June 3, the same day SpaceX announced its $135 offering price, Tesla closed at roughly $424 per share.

At that price, Musk’s Tesla holdings were worth approximately:

· $175.2 billion from common shares

· $121.8 billion from stock options

for a total of approximately $297.0 billion.

The Trillion-Dollar Calculation

Musk also owns stakes in several other companies, including The Boring Company and Neuralink.

Ignoring those holdings entirely and counting only SpaceX and Tesla, his net worth would be:

Asset

Value

SpaceX

$819.3 billion

Tesla

$297.0 billion

Total

$1.116 trillion

In other words, Musk crosses the trillion-dollar threshold even before assigning any value to his holdings in xAI, Neuralink, or The Boring Company.

What If SpaceX Is Worth More?

One could argue that using the SpaceX offering price is too conservative.

Around the same time, SpaceX shares were trading at roughly $180 in secondary markets. Using that price instead, Musk’s SpaceX stake alone would be worth approximately $1.092 trillion.

Under that valuation, his ownership in SpaceX by itself would exceed one trillion dollars.

Updating for Recent Market Moves

The calculations above are based on market prices as of June 3.

Since then:

· Tesla’s stock price has fallen from roughly $424 to around $380.

· Secondary-market prices for SpaceX shares have declined from roughly $180 to around $160.

These moves reduce the value of Musk’s Tesla holdings by approximately $31.6 billion and value his SpaceX stake at roughly $971.0 billion.

Yet even under these more conservative assumptions, Musk’s fortune remains near the trillion-dollar mark.

How Much Is One Trillion Dollars?

It is difficult to develop an intuition for how much money one trillion dollars really is.

Consider the following thought experiment.

Suppose you could spend $1,000 every hour, twenty-four hours a day, seven days a week.

How long would it take to spend one trillion dollars?

The answer is one billion hours.

One billion hours is roughly 114,000 years—far longer than the history of human civilization itself.

That is the scale of wealth we are talking about.

Disclaimer

This piece reflects my personal estimates based on publicly available filings, prospectus disclosures, and market prices. Some calculations involve simplifying assumptions and should be viewed as illustrative rather than precise measurements of Elon Musk’s net worth.

The main points from that video were as follows:

1. Walmart and Costco are trading at extraordinarily high valuations.

As of April 29, 2026, Walmart traded at a P/E ratio of 47, while Costco traded at 52. For context, Walmart’s historical average P/E ratio is roughly 20. Even allowing for some structural changes in the business, today’s valuation appears remarkably high by historical standards.

2. The “Walmart is becoming a tech company” explanation is not fully convincing.

One common argument is that Walmart deserves a higher multiple because it is evolving into a technology-enabled platform, much like Amazon. However, this explanation raises an obvious question: if Walmart is becoming more like Amazon, why does Walmart trade at a substantially higher P/E ratio? At the time of the video, Amazon’s P/E ratio was about 31, far below Walmart’s 47.

3. Walmart remains one of the biggest valuation puzzles I have encountered.

I teach both Behavioral Finance and Equity Valuation at the University of Florida. Since at least Spring 2025, I have told my students that Walmart’s valuation is one of the biggest stock market puzzles I have seen.

Back then, Walmart traded at approximately 35 times earnings. My expectation was not necessarily that the stock would decline, but that the valuation multiple would eventually normalize. Instead, over the following year, the P/E ratio expanded further. As a result, I am even more puzzled today than I was then.

Fast forward to today. Additional puzzles have emerged.

4. The puzzle persists today.

Updating the numbers to June 9, 2026, Walmart’s P/E ratio has declined from 47 to 42, while Amazon’s has fallen from 31 to 29. Although Walmart’s valuation is no longer as extreme as it was in April, it remains dramatically above its long-run historical average.

5. Walmart appears to be trading as a hedge against the broader market.

An even more curious development is Walmart’s recent behavior during market selloffs.

For example, at approximately 3:15 p.m. on June 10, 2026, SPY was down 1.4%, while Walmart shares were up 0.83%. This was not an isolated incident. Over the past several trading sessions, Walmart has frequently outperformed during periods of market weakness.

Historically, Walmart’s beta has been around 0.6. In other words, when the S&P 500 rises by 1%, Walmart typically rises by about 0.6% as well. Walmart has generally behaved as a lower-volatility equity, not as a negatively correlated asset.

Yet over the past several trading days, investors seem to have treated Walmart almost as a defensive hedge against broader market weakness. That strikes me as unusual and potentially important.

6. The valuation is even more surprising given today’s interest-rate environment.

Perhaps the most striking aspect of this story is that it is unfolding during a period of relatively high interest rates.

As of yesterday (June 9, 2026), the yield on the 10-year U.S. Treasury is approximately 4.5%, while the 30-year Treasury yields around 5%. Traditionally, higher interest rates put downward pressure on equity valuation multiples because future cash flows are discounted at a higher rate.

This effect is usually strongest for long-duration assets and growth stocks, but it should also matter for mature, slower-growing companies such as Walmart. Yet despite long-term Treasury yields hovering around 4.5% to 5%, Walmart continues to trade at more than double its historical average P/E ratio.

If Walmart were trading at 20x earnings, I would not find that surprising. At 42x earnings in today’s interest-rate environment, I do.

The high-interest-rate environment makes the valuation puzzle even harder for me to understand.

I do not have a satisfying explanation for any of this.

Why is Walmart trading at more than double its historical valuation multiple? Why does the market assign Walmart a higher earnings multiple than Amazon? Why has Walmart recently behaved almost like a hedge during market declines? And why is all of this happening when long-term Treasury yields are near 5%?

I would greatly appreciate hearing from readers who have a theory. What am I missing?

Disclaimer: This piece reflects my personal views. Some of the figures are back-of-the-envelope estimates intended to illustrate broad patterns rather than provide audit-level precision. Read—and invest—at your own risk.

One question: how much of this trend is structural rather than speculative?

Over the past four decades, the U.S. retirement system has shifted from defined-benefit pensions to defined-contribution plans such as 401(k)s. As millions of workers automatically direct retirement savings into equity-heavy portfolios, a growing share of household wealth is continuously funneled into the stock market.

If so, today’s record equity allocation may reflect not only investor optimism, but also a fundamental transformation in how Americans save for retirement.

But if you peel back the curtain and look at the actual anatomy of these corporate earnings, a bizarre structural distortion emerges.

A massive chunk of this quarter’s blockbuster “growth” didn’t come from selling more software, shipping more microchips, or delivering more packages. Instead, it came from an accounting rule that forced massive, illiquid “paper gains” onto the income statements of tech giants.

In Q1 2026 alone, just three companies—Alphabet, Amazon, and Nvidia—reported a staggering $69.2 billion in non-operating windfall under their Other Income and Expenses (OI&E) lines. When you run the macro numbers, this single accounting phenomenon artificially inflated the entire S&P 500’s quarterly earnings by about 12%.

Understanding OI&E

In standard corporate accounting, Other Income (Expense), net is typically a sleepy, neglected corner of the financial statement. It is designed to capture items strictly outside a company’s core operations—things like interest earned on cash reserves, foreign exchange fluctuations, or minor investment tweaks.

Historically, this line item was a rounding error. Today, it’s an engine of massive earnings distortion.

The culprit is an accounting standard known as ASU 2016-01. Under GAAP rules, companies are required to measure their equity investments in both private and public companies at “fair value” at the end of every single quarter.

· Up-Round Markups: If a startup they invested in raises a new private funding round at a higher valuation, the corporate backer must mark up the value of their shares on their own balance sheet.

· Income Statement Impact: Crucially, that imaginary, unrealized “paper profit” must be funneled directly through the income statement under the OI&E line.

Because we are currently riding an unprecedented wave of capital concentration into private AI infrastructure, late-stage tech startups, and secondary markets, these private valuations have gone completely parabolic.

The Big Three Windfalls

To see how extreme this trend has become, look at the literal billions that hit the books of Big Tech this quarter (all dollar figures in billions):

Let’s be entirely clear about what these numbers mean: 60% of Alphabet’s, 51% of Amazon’s, and 27% of Nvidia’s reported net income this quarter came from non-operating, illiquid paper gains. The only issue of these earnings is that their “quality” is lower than operating earnings, in the sense that we do not expect them to reoccur.

The Macro Math of the S&P 500

To put a $69.2 billion non-cash surge into perspective, we have to look at the total earnings footprint of the entire index.

Annualized total earnings for the S&P 500 are currently projected to hover around $2.3 trillion. If we split the difference at a baseline of $2.3 trillion for the full year, a typical, clean quarter should yield roughly one-fourth of that total:

$2.3 Trillion / 4 = $575 Billion in Quarterly S&P Earnings

Now, let’s look at the mathematical leverage that just three companies’ non-operating paper profits exerted over the entire index’s bottom line:

$69.2 Billion (Big Three OI&E) / $575 Billion (Total S&P Q1 Earnings) = 12.0%

The Takeaway: More than 12 cents out of every single dollar of profit reported across all 500 of America’s largest companies this quarter was an illiquid, non-operating mark-to-market adjustment from just three balance sheets. Such an effect inflates the earnings by 12% but inflates the earnings growth by about 75% (from 16% to 28%). Without the OI&E lines, the Q1 2026 earnings growth rate will not be very different from the 5-year average of 16%.

The Danger of a “Circular Valuation” Loop

For Substack readers and macro investors, this creates a highly volatile, circular feedback loop that investors should be watching closely:

· 1. Public Multiples Expansion: Public markets trade at premium, heavily driven by the AI narrative.

· 2. The Private Premium: High public multiples justify sky-high private valuations for venture-backed AI startups during private funding rounds.

· 3. The Earnings Mirage: Big Tech marks up those private holdings, inflating their own public GAAP earnings via OI&E.

· 4. The Justification: Wall Street looks at the “blowout” public GAAP earnings, declares tech valuations are fully justified by earnings growth, and bids the public stock prices even higher.

Please note that the underlying operating health of these businesses remains exceptionally robust—both Amazon’s AWS and Google Cloud grew very fast. They don’t need to fake it.

But by letting the volatile, illiquid world of venture capital markups dictate public net income lines, the market has introduced a massive blind spot. If the private venture market chills out later this year, or if an AI scale-up undergoes a down-round, these exact same accounting rules will force billions in non-operating losses down the index’s throat.

But a more likely scenario is that we should not expect these OI&E earnings to reoccur. I am not sure whether the market has made the distinction between operating earnings and OI&E. As discussed above, without the OI&E lines, the Q1 2026 earnings growth rate will not be very different from the 5-year average of 16%.

Disclaimer: This piece reflects my personal speculative take. Some of the aggregate numbers are baseline estimates meant for illustrative mapping rather than hyper-precise auditing. Read (and invest) at your own risk.

First, Owning SpaceX May Be the Closest Thing to Apocalypse Insurance.

SpaceX may be the ultimate tail-risk hedge. In fact, it could be humanity’s only meaningful hedge against an existential catastrophe. If Earth ever becomes sufficiently hostile to human life, the most valuable asset in the world will not be gold, farmland, or Bitcoin—it will be a ticket off the planet. Demand for SpaceX seats would become effectively infinite, and ticket prices would rise accordingly. In that world, owning SpaceX stock today may be the closest thing to buying insurance against the collapse of civilization. The irony, of course, is that when everyone wants a seat on the rocket, only the people who bought the rocket company will be able to afford one. So, are you going to hedge, and how much are you willing to pay?

Second, SpaceX's IPO could sink Tesla.

SpaceX’s most formidable competitor is not Blue Origin, Rocket Lab, or any other aerospace company. It is Tesla. The two firms are not competing for customers; they are competing for shareholders/believers.

Both companies are controlled by Elon Musk. Both trade—or would trade—at extraordinarily high valuations. Both derive much of their value from narratives about the future, technological optimism, and investors’ faith in Musk’s ability to bend reality. Tesla’s market capitalization exceeds $1 trillion, while SpaceX is valued at roughly $2 trillion in private markets. To a significant extent, investors are buying exposure to the same underlying asset: the Musk vision.

This has an interesting implication. Miller (1977) argues that when investors disagree and short-selling is constrained, asset prices can be elevated because optimists set the price. But introducing additional supply reduces this effect, more so if investors’ opinions are more divergent. If Tesla and SpaceX are viewed as close substitutes in the market for Musk-powered dreams, then a SpaceX IPO would effectively increase the supply of “Musk equity.” And when the supply of dreams increases, the price of dreams tends to fall, because now you need the less optimistic investor to participate.

In other words, the greatest threat to Tesla’s valuation may not come from Chinese EV makers, autonomous driving failures, or declining margins. It may come from Elon Musk selling investors another, even bigger dream.

These are merely a few speculative thoughts on a SpaceX IPO. Read them at your own risk.

The S&P 500’s P/S ratio has reached a record high, standing roughly two standard deviations above its long-run average. By contrast, the P/E ratio is currently around 32—elevated by historical standards, but still below its all-time peak.

The difference between these two valuation measures points to one conclusion: corporate profit margins are exceptionally high. Investors are paying record prices relative to sales, while strong profitability has kept earnings growing fast enough to prevent the P/E ratio from reaching comparable extremes.

This raises an important question: Are today’s profit margins sustainable, or is the market implicitly assuming that they will remain unusually high for years to come?