Commonsense Capital

Micron trading at 21x sales. That's more than double its dot-com bubble peak of 10x.

Just let that sit for a second.

We're talking about a semiconductor company — cyclical, capital-intensive, commodity-ish products — being valued like it's a SaaS unicorn with 90% gross margins.

$MU isn't some magical AI story immune to supply/demand cycles. It makes memory chips. The industry has a long, painful history of boom-bust oversupply disasters.

Maybe this time is different. Maybe AI demand is so insatiable that Micron prints cash forever and justifies 21x sales.

Or maybe we're watching the same movie again, just with better special effects.

Valuation doesn't tell you what happens tomorrow. But it tells you a lot about what's already priced in — and what happens when expectations meet reality.

$AMAT is trading at 16x+ sales — higher than the dot-com peak in April 2000 (15x).

When a semiconductor equipment company is more expensive than it was during the biggest tech bubble in history, that's not a bullish signal. That's a flashing red light.

Valuations this stretched don't end well. They never do. The only question is timing.

SpaceX trading at 125x sales. Let that sink in.

For context: $NVDA, the poster child of AI mania, sits at 20x. $TSLA at 15x. The entire $SPX at 3.6x.

SpaceX is priced like it's already colonized Mars and monetized the trip.

I get it — Starlink has potential, reusable rockets are revolutionary, Elon is Elon. But at some point, valuation has to connect to actual cash flows, not just vibes and future promises.

This is what happens in late-cycle euphoria: people stop asking "what's it worth?" and start asking "how high can it go?"

History says badly. Every time.

You don't have to short it. You don't have to hate it. Just don't pretend the math makes sense.

The $AI discourse is schizophrenic.

One minute people are breathlessly predicting AGI/ASI will arrive any day now and fundamentally reshape civilization.

The next minute they're panicking because Mistral is 9 months behind OpenAI on some benchmark.

Pick a lane.

If you actually believe we're on the verge of recursive self-improving superintelligence, then who's ahead in the model race today is utterly irrelevant. It's like worrying about which telegraph company has better infrastructure in 1875 when you think the internet is coming in 1876.

But if you're sweating the competitive positioning of specific AI labs, then you clearly don't believe your own hype about imminent transformative AI.

Most of the $AI investment thesis is built on this contradiction. People want the valuation multiples that come from "this changes everything" while simultaneously trading on incremental product cycles like it's just another software category.

Can't have it both ways. Either this is the most important technology in human history and current market share means nothing, or it's a normal innovation cycle and we should value these companies like normal companies.

My guess: it's somewhere in between, which means both the doomers and the hype merchants are wrong.

SpaceX at $2.3T would be the 6th largest US company by market cap — but 198th by revenue.

That's a valuation stretched so far beyond current fundamentals it makes 1999 tech stocks look conservative.

Investors are pricing in exponential growth that's literally never happened at this scale. Not Amazon. Not Apple. Not Google.

Maybe they're right. Maybe Starlink dominates global internet, Starship revolutionizes logistics, and Mars colonies become reality.

Or maybe this is just what peak euphoria looks like when you combine a charismatic founder, genuine technological progress, and zero interest in boring things like cash flow.

History says when revenue is 200 spots behind market cap, you're not investing — you're speculating on a miracle.

Nothing wrong with that. Just know what game you're playing.

Back in 2017, I wrote about how wrong the tech bubble calls were.

It's 2025. Still waiting for that bubble to pop.

Almost a decade of people being early, staying early, and missing one of the greatest wealth creation periods in history.

The lesson isn't that bubbles don't exist. It's that calling tops is a losing game. Time in beats timing.

Both $BTC and gold down YTD in 2026 — haven't seen that combo before.

Gold's barely red (-3%), but bitcoin's gotten hammered (-27%). When your "inflation hedge" and your "digital gold" both lose together, it's usually a sign that liquidity is tightening and real rates are biting.

This isn't some philosophical debate about store of value anymore. It's just math: when cash yields something and risk appetite fades, speculative assets get sold first, hard assets get sold slower.

Historically, gold and bitcoin don't move in lockstep. When they both go down, it's worth asking what's actually driving flows — not what the narrative says should be driving flows.

Tech just had its biggest 9-week run vs the S&P 500 in history — 28% outperformance. Bigger than the dot-com mania in late 1999/early 2000.

Let that sink in.

When you're beating records set during the most famous bubble in modern market history, you're either witnessing something truly revolutionary... or you're watching people forget how gravity works.

I'm not saying sell everything. I'm saying check your position sizes and ask yourself: what am I actually betting on here? And what happens if the narrative shifts even a little?

$XLK

The Economist nailed it: "In global trade, limitation is the sincerest form of flattery."

China's export machine keeps growing. The rest of the world? Throwing up tariffs and restrictions.

It's the oldest playbook: flood markets with cheap goods, build dominance, provoke protectionism. Nobody wants to admit they can't compete, so they just block the door.

History says this ends one of two ways — either China pivots to domestic consumption (they've been "pivoting" for 15 years), or trade wars drag on until something breaks.

Either way, investors need to stop pretending globalization is a one-way street. Supply chains are fragmenting. Costs are rising. The era of frictionless trade is over.

Plan accordingly.

Markets ripping today — $NASDAQ up 2.8%.

Nice bounce, but zoom out. One green day doesn't change the trend. Don't mistake relief for reversal.

If you're tempted to chase because you feel FOMO, ask yourself: would you have bought yesterday when it was red? If not, you're trading emotions, not conviction.

Stay disciplined. The best investors don't react to every move.

Three weeks into 2026 and we're watching a textbook rotation unfold:

• International & EM stocks outpacing the $SPX
• Value beating growth
• Small/mid caps crushing large caps
• Mag 7 in the red

This is what mean reversion looks like when it actually shows up. After years of U.S. mega-cap tech dominance, the pendulum is swinging hard the other way.

Does it stick? Maybe. Maybe not. But this is why diversification isn't just a textbook concept — it's insurance against exactly this kind of regime shift.

If you spent the last three years convinced "there is no alternative" to the Mag 7, you're learning an expensive lesson right now. Markets don't move in straight lines forever, and what worked yesterday doesn't always work tomorrow.

Don't chase. Don't panic. Just remember: the best performers rarely repeat, and the worst performers rarely stay worst. That's not a hot take — that's just how markets work.

Hosted our first Odd Lots trivia night in Hong Kong last week. Good turnout, met a bunch of listeners, everyone seemed into it.

Then Travis Lundy took over as guest host for a round and the entire room went dead silent. Not sure what happened there but the vibe shift was... notable.

The "shrinking middle class" headline everyone panics about?

Look closer: a big chunk of that shrinkage is people moving UP into upper-middle-class brackets. American households are actually getting richer.

Yes, the rich are getting richer faster. Yes, wealth inequality is real and complicated.

But the narrative that the middle class is collapsing into poverty is lazy. Upward mobility is still happening — just unevenly.

Context matters. Don't let headlines do your thinking for you.

The $SPX has been anything but boring this decade.

2020: +18.4%
2021: +28.9%
2022: -18.2%
2023: +26.2%
2024: +24.9%
2025: +17.8%
2026: +10.7% (YTD)

That's 15.7% annualized. Wild swings, no calm years, yet the compound return keeps humming along.

This is what equity markets actually look like — volatile, uncomfortable, but rewarding if you stick around. The problem isn't the volatility. It's that most people can't handle it psychologically and bail at exactly the wrong time.

If you're sitting here complaining about market risk or waiting for "the right entry," you've already missed seven years of this. The cost of being clever is almost always higher than the cost of being patient.

People wildly underestimate how much of career success is just dumb luck.

Right place, right time. Meeting the right person at a conference. Getting laid off from a dying company before you wasted 5 more years there. Your boss leaving and you getting promoted because someone had to fill the seat.

The best investors I know? Half of them admit their first big win was an accident. Bought something for the wrong reasons, held it because they forgot about it, made 10x.

Doesn't mean skill doesn't matter. It does. But luck gets you in the door. Then you need the sense to recognize it and not blow it.

Most people do the opposite — they get lucky once, think they're geniuses, lever up, and give it all back plus interest.

The real skill isn't making money when you're lucky. It's not losing it when you're not.

New research pokes holes in the "AI doctor" hype.

LLMs crush medical licensing exams in controlled settings, but accuracy collapses when you rephrase the same question differently. Classic overfitting to benchmark patterns, not actual understanding.

This matters beyond medicine. We're seeing similar brittleness across domains — models that look brilliant on static tests but fall apart when real users ask questions in unpredictable ways.

Before you trust any AI benchmark score at face value, ask: what happens when you stress-test it with real-world messiness? Usually the answer is: performance craters.

Benchmarks measure what we can easily measure. Reality measures what actually matters.

SpaceX just raised another massive round. Impressive, sure. But here's the real question nobody's asking yet:

Can US capital markets actually keep up with what's coming?

We've gotten used to seeing these giant funding rounds — $10B, $20B, $50B valuations like it's nothing. The machine works. Money flows. Everyone cheers.

But demand for long-term capital is about to spike hard across multiple fronts:
• Defense/aerospace buildout
• Energy transition infrastructure
• AI compute and data centers
• Reshoring manufacturing
• Climate adaptation

All capital-intensive. All long-duration. All competing for the same pool of patient money.

Meanwhile, traditional sources of that patient capital — pension funds, insurers, sovereign wealth — are already stretched thin, dealing with higher rates, duration risk, and their own liability mismatches.

So yeah, SpaceX got funded. Great. But can the market do that 50 more times in the next few years without breaking something?

This isn't a doom prediction. It's a capacity question. And it's going to matter a lot more than people think.

3.5 months into a major war and $BTC Brent's at $83.

Markets don't care about geopolitics the way people think they do. They care about supply, demand, and whether the actual flow of oil gets interrupted. Drama ≠ disruption.

This weekend's headlines might change that equation. Or they won't. Either way, oil prices tell you more about reality than the news cycle does.

Watch what trades, not what trends.

Nasdaq futures up nearly 2% overnight.

This is what happens when everyone's been positioned for doom and gloom. One decent signal and the snap-back is violent.

Reminder: the market doesn't care about your narrative. It cares about positioning, flows, and who's forced to cover.

Don't chase. Don't get euphoric. This could be a real turn or just another bear market rally. Nobody knows yet.

Stay patient. Let the picture develop.

New Odd Lots episode just dropped — we sat down with Carmen Li from Compute Exchange to talk about something genuinely novel: compute futures launching on the CME.

Think about it: we've had oil futures, wheat futures, currency futures for decades. Now we're getting standardized contracts for computing power. The infrastructure layer of the AI economy is becoming tradable.

What's interesting here isn't just the product itself — it's what it signals. When you can hedge compute costs with CME-listed futures, you're watching a new asset class get financialized in real time. Cloud computing costs move from operating expense line items to something you can price, hedge, speculate on.

For context: companies are burning billions on AI compute. Uncertainty around GPU availability and pricing is a real business risk now. A liquid futures market could actually solve a problem, not just create new ways to gamble.

Worth a listen if you want to understand how markets evolve when new scarcities emerge. This isn't crypto vaporware — this is the CME, actual infrastructure, real counterparty risk management.

Compute is the new oil? Maybe. But at least now there's a curve.