Why Bitcoin WINS No Matter What Happens to Inflation

ELI5/TLDR

Pompliano talks to hedge fund manager Jordi Visser about why Bitcoin and software stocks are quietly divorcing. The argument: software is dying because AI strips its margins, Bitcoin wins on both sides of the inflation/deflation seesaw, and the real money this cycle is in things you can physically touch — chips, chemicals, servers, copper. Visser also pitches Bitcoin allocation as a probability bet: if there’s a 30% chance the fiat system gets rewired by AI in 20 years, your portfolio should reflect 30%, not 1%.

The Full Story

The decoupling

Visser opens with a chart fact: this month, almost every winning name in the IGV software ETF was crypto-linked. Software bellwethers (IBM, ServiceNow) reported fine earnings but the market punished the commentary. The reason is a slow realization in Silicon Valley that AI agents have removed the terminal value of pure-software companies — meaning analysts can no longer plug a “company exists in 10 years” assumption into their models.

Bitcoin miners, meanwhile, became valuable for a non-Bitcoin reason: they have power, racks, and cooling infrastructure that can be repurposed for AI compute. Compute is now in shortage. So miners caught a bid first, and Bitcoin followed because a healthier mining ecosystem helps the network.

The two-sided trade

Pompliano’s pet thesis is that Bitcoin works in both inflationary and deflationary regimes. Visser agrees and refines it via Jeff Booth’s frame: a credit-based fiat system always has deflationary pressure underneath because innovation makes things cheaper. When that pressure hits ordinary people — no housing, no upward mobility, jobs that don’t match degrees — it generates anger, and anger generates Bitcoin demand.

The other side is government response. Deflationary pressure forces stimulus. Stimulus plus commodity bottlenecks creates negative real rates: short rates pinned low because the underlying economy is weak, headline CPI running hot because physical goods are scarce.

“To have negative real rates, you have deflationary pressures which are keeping short-term rates low, and you have inflationary pressures which are keeping CPI high.”

Bitcoin sits at the intersection. It benefits from the inflation side directly and from the deflation side indirectly, because when growth assets like hyperscalers stop working, investors needing 40-year compounding have fewer places to go.

The scarcity portfolio

Visser walks through the thematic baskets he’s been writing about for institutional clients. The unifying logic is shortage. He’s a self-described “scarcity trader” trained originally as a horse-racing handicapper, which means he reads where the smart money is positioned the way a handicapper reads a tote board.

Semiconductors — Now the largest industry group in the S&P 500 at 17% weight. Of the roughly $10 trillion in semi market cap, three-quarters sits in Nvidia, Broadcom, and Micron. The early winners were data-center GPU plays. Since November, the shift has been toward inference and agents, which is why memory (Micron) and connectivity names (Marvell, with silicon photonics) have caught up.

Chemicals — A non-obvious AI play. Chemicals are PMI-sensitive and also feed semiconductor packaging and Corning’s optical fiber tubing. US chemical producers benefit because they run on cheap domestic natural gas while overseas competitors run on oil. Visser frames it as the start of a chemicals bull market.

Whole Rack — Servers, memory, CPUs, the full stack a Morgan Stanley-type firm needs when it decides for privacy reasons to host AI on-premise instead of in someone else’s cloud. Dell and HPE have broken out on this.

The auto upgrade cycle — Annual US car sales have been stuck around 16 million since the 1970s. Visser’s pitch: it’s not unit growth that matters, it’s content-per-car. If 1 million of this year’s cars have AI components and next year’s number is 4 million, you’ve created a 300%+ growth market inside a flat one. Jensen Huang has framed the eventual replacement of the US fleet as an $85 trillion recycling.

The Contracted Annual Run Rate problem

A useful detour into one of the era’s accounting tricks. Annual Run Rate (ARR) was already loose — take one month’s revenue, multiply by 12. Now there’s “Contracted Annual Run Rate” (CARR), which lets a company headline a number based on year three of a back-loaded contract.

The setup: an AI company signs a customer for a 3-year deal. Year 1 costs $100k, year 2 costs $1M, year 3 costs $3M. The customer can cancel after 12 months. The vendor reports $3M in CARR even though only $100k is locked in and the customer can walk.

“Unsophisticated investors may not realize that I really only got 100k or they could get out after a year.”

Visser’s broader point: this trick exists because no one knows what the right valuation is for a SaaS company anymore. With exponential change, terminal value is undefined, so both sellers and buyers retreat to short-dated, optionality-heavy contracts. The CARR number is the cosmetic cover for that uncertainty.

The 20-year fiat question

The conversation drifts into the Visser worldview. He’s a probability thinker. If you believe humanoid robots, radical life extension, and an AI productivity wave together have a meaningful chance of breaking the fiat system in 20 years — say 30% — then your Bitcoin allocation should look like 20-30%, not 1-5%. He’s pitching this to RIAs and wealth managers as a position-sizing argument, not a price-target argument.

“If I think that’s a 30% chance of happening, and in my net worth, if I have 5% in Bitcoin or 1% in Bitcoin, that’s the wrong number. The right number is 30%, it’s 20%.”

The political subplot: Visser sees the Mamdani election (the show treats this as having happened, with 70% support in the East Village) as confirmation of Strauss-Howe’s Fourth Turning thesis. The generation born after the Great Depression treated debt as bad. Their grandchildren do not. When that generational memory finishes turning over, the political possibilities shift, and so does the cost of capital.

The closing detour

The last 10 minutes drop the macro thread for a discussion of stoicism, Marcus Aurelius, and Visser’s experience losing his best friend on 9/11 — which is what pushed him out of Morgan Stanley. It’s the Pompliano podcast house-style ending: a personal coda that has nothing to do with the trade.

Key Takeaways

Software stocks and Bitcoin are decoupling — software losing its terminal value to AI agents, Bitcoin migrating into the PMI-sensitive scarcity bucket
Visser expects negative real yields: headline CPI above 4%, 3-month bills below 4%, driven by physical-goods bottlenecks (memory, CPUs, chemicals, freight)
Five scarcity baskets: semis, chemicals, whole rack (servers), power, optical/photonics — all downstream of the AI buildout
The smart car upgrade cycle is the under-appreciated semi growth story — content-per-car, not unit growth
“Contracted Annual Run Rate” is the new accounting tell — companies headline year-3 numbers from contracts the customer can exit after year 1
Visser’s allocation pitch: size Bitcoin as a probability of fiat-system disruption, not as a tactical trade. If you think there’s a 20-30% chance, allocate accordingly

Claude’s Take

Pompliano runs a maximalist show, and the title “Why Bitcoin WINS No Matter What Happens to Inflation” is the giveaway — that’s an unfalsifiable framing. If Bitcoin wins in inflation and Bitcoin wins in deflation, then Bitcoin wins always, and the thesis has no kill condition. That’s a sales pitch, not a model. Worth flagging up front.

That said, Visser is the more interesting half of the conversation. The decoupling observation is real and verifiable — you can pull the IGV ETF and see which names have led this month. The “no terminal value” framing for SaaS is a sharp way to articulate something the market is genuinely struggling with: how do you DCF a company when the technology underneath it might be obsolete in 18 months? That problem is not hype, it’s a real valuation crisis.

The CARR breakdown is the most useful five minutes of the show. Anyone doing diligence on AI-adjacent SaaS names should be reading their disclosed revenue numbers with that filter on. The trick of headlining year-3 contract value while the customer has a 12-month exit is the kind of thing that ends in writedowns.

Where Visser stretches credibility is the 30% probability bet. Probabilistic position sizing is a respectable framework, but the inputs here — humanoids, radical life extension, fiat-system collapse — are estimates with no real base rate. He’s essentially saying “if you believe my unverifiable forecast, then size accordingly.” A discipline professional gambler would not bet on a horse race where he assigned the odds himself.

The semiconductor and chemicals analysis is the strongest section. The “$10T market cap, three names hold $7.5T” stat is concrete and useful. The chemicals-as-AI-input thesis is non-obvious enough to be worth chasing. The auto content-per-car argument is the kind of mental model that travels well into other sectors — it’s not unit growth, it’s value-per-unit growth.

Six out of ten. Useful for the scarcity-trade frame and the CARR tell, ignore the title’s promise.