The list landed on October 25. No year attached. Eleven claims, and when I ran them line by line the way I'd run a wallet cluster, ten were opinions and one was a fact — that a podcast segment happened. Zero transaction hashes. Zero revenue multiples. Zero time-stamped falsification conditions.
I built my reputation on speed. In November 2022 I had Alameda's outflows mapped to three shell entities before the bankruptcy docket was even searchable, and that night taught me something permanent: verification speed beats breaking speed. So I don't read a list like this for what it says. I read it for what it's structured to hide. What this one hides is a single trade wearing four costumes.
Anthony Pompliano is a known quantity — investor, newsletter operator, podcast host, one of the loudest long-cycle Bitcoin voices since 2017. His October commentary set out a twenty-year thesis: Bitcoin, gold, and land outperform; AI has enormous runway and specific AI companies are severely undervalued; over the next five years venture capital and public markets converge while capital concentrates into a handful of quality companies, pushing private equity valuations higher. Then the part every aggregator clipped: young investors should hunt where Wall Street hasn't dug — private equity, prediction markets, compute markets, data brokerage. And a closer, delivered flat: Tesla and SpaceX merge.
Why it matters now isn't the Bitcoin call. Nobody in a bear market needs another twenty-year Bitcoin call. What matters is that this specific bundle of four sectors is being handed to retail at the exact moment each one walks into its regulatory reckoning. That's a timing signal, not an allocation signal.
Let me take them in order of technical debt.
Prediction markets are the cleanest business model and the dirtiest legal question in the entire list. They generate real fee revenue — no token subsidy, no liquidity mining, an actual product people pay for. That's rare and I respect it. But the core dispute isn't technical. It's whether an event contract is a swap, a commodity, or a wager. The CFTC has taken the federal view; multiple state gaming regulators have claimed the opposite jurisdiction. Smart contracts don't file for gaming licenses. The entire cost structure of the business changes depending on which side wins — a gaming-license regime and a derivatives regime are not the same company. Anyone buying this sector off a podcast should be tracking the docket, not the chart.
Compute markets I know from the inside. I've spent audit cycles on DePIN compute networks, and the structural problem is verification at three layers: hardware authenticity (is that really an H100, or a relabeled A100 with spoofed telemetry), execution authenticity (did the node actually run the job, or did it proxy to a hyperscaler and pocket the spread), and latency penalty across geographies. The token subsidy hides all three, because when the reward per unit of work exceeds the cost of faking it, rational operators fake it. Render, Akash, io.net all wrestle with this. Strip the emissions and you find out how much real inference demand exists. My prior, from watching this film since 2020: less than the market cap implies.
Data brokerage is where the advice turns genuinely strange. Selling personal data on-chain inside a GDPR and CCPA environment isn't a neutral technical choice — it's a values position running directly against every major privacy regulator on earth. The chain makes the transfer auditable, which is exactly what makes the compliance failure auditable too. You aren't removing the intermediary. You're manufacturing permanent evidence.
And AI agents in finance — the one genuinely interesting thread. An agent that moves money needs four primitives: programmable authorization boundaries, auditable identity, revocable permissions, and a small-value high-frequency settlement rail. That's account abstraction plus session keys plus intent architecture plus stablecoin rails. Real design space, and 2024-2025 produced real work here. But "AI reshapes finance" without naming the layer isn't analysis. Execution layer, intermediation layer, clearing layer — each carries a different order of magnitude of regulatory resistance. He named none of them.
Now the pattern. All four share one trait: demand already proven by TradFi, token model unproven, and the current moat is a compliance cost gap. The excess return in this list isn't technological. It's the price of regulatory ambiguity — and ambiguity has a half-life.
Here's the paragraph the aggregators dropped, the only one with real information density: VC and public markets converging, capital concentrating into fewer quality companies, private equity valuations pushed up.
Read those three together and a contradiction surfaces that nobody flagged. Concentration raises the valuations of the survivors. Raised valuations compress forward returns for anyone buying at that price. The end state of concentration is not "winners return more" — it's "winners are priced for perfection." Look at the 2024-2025 AI primary market: top-tier names raising at multiples that demand a decade of flawless execution. That's the mechanism running live, in front of us.
And convergence is not democratization. When private assets get wrapped into public vehicles, the liquidity risk and the valuation risk migrate to the retail holder. It's risk transfer sold as access. The exit liquidity was already gone long before the wrapper reached the brokerage app.
The tell is the Tesla-SpaceX line. Untradeable, unfalsifiable, unforgettable — which is precisely its function. It's an anchor that makes four sector picks read as moderate by comparison.
So here's what I'm watching, and it isn't price. It's the prediction-market jurisdiction ruling, the GPU export control schedule, and the first enforcement action against an on-chain data broker. Those three events reprice the entire list. Twenty years is a horizon chosen so it can never be wrong. Volatility is just velocity without direction — and a thesis with no falsification date has no direction at all.