Jordi Visser — Scarcity vs. Abundance Investment Thesis

Source: AI Is Destroying Every Moat Except Bitcoin | Jordi Visser, Bankless, April 27, 2026.

The Framework: Scarcity vs. Abundance

Visser's unifying lens for every position he holds: own scarcity, exit abundance.

Abundance (exit)	Scarcity (own)	Investment Implication
Software / SaaS — anything built on code	Bitcoin, silver, copper, DRAM	AI commoditizes code; atoms remain finite
Corporate earnings per share (S&P flat 10yr)	Physical AI infrastructure — chips, memory, energy, minerals	Productivity flows to capital and entrepreneurs, not to existing corporations
Knowledge work / professional services	Compute shortage bottleneck	"Get out of anything abundant; only invest in things that are scarce"
Private credit (SaaS-collateralized)	Hard assets / store-of-value assets	Terminal value of SaaS collapses; debt backed by it collapses with it

Investment Thesis #1: Bitcoin Is the Last Surviving Moat

Argument: AI destroys the terminal value of every asset class built on code or replicable intelligence — software, artwork, professional credentials, data. Bitcoin is the single exception: it is physically constrained (energy, chips), socially chosen (gold, religion, Bitcoin — the three moats that survived), and structurally scarce. Visser's thesis is that the same force (AI) that is destroying software moats will eventually drive capital toward the one thing AI cannot replicate.

"AI is basically to all of these companies built on code as quantum is to Bitcoin. It makes people question the value going forward of the asset."

Most traditional investors see Bitcoin and AI as unrelated or competing. Visser argues they are the same trade: AI abundance destroys all soft stores of value, making the hard scarcity of Bitcoin more valuable, not less.

Trigger: CPI year-over-year exceeds 3-month T-bill yields + Fed holds or eases → the regime in which 100% of Bitcoin's historical returns have occurred; "247% annualized" in this quadrant.

Names: Bitcoin (BTC), MicroStrategy (MSTR) as the largest public Bitcoin proxy.

Investment Thesis #2: SaaSapocalypse — Terminal Value Collapse for Software

Argument: Software companies are not being disrupted by lost revenue — their current earnings are fine. They are being disrupted by terminal value compression. If AI can generate software by voice command today, the long-run cash flows of Salesforce, Adobe, and the entire SaaS universe approach zero. This is the same dynamic as the discount rate rising to infinity: investors with inside knowledge (Silicon Valley) are already exiting software, which explains weak software stock performance despite strong current earnings. Private credit, which is 25–50% SaaS-collateralized, is the next shoe.

"The disruption from AI will start to be built in... the majority of money — if you wanted to figure out who owns most of the stocks in the US, think about it, Silicon Valley owns a really high portion of them. If they know the disruption is coming, do they want all their money sitting in something that's going to be disrupted?"

Wall Street calls software earnings "fine." Visser says the stock market is correctly discounting future destruction, not incorrectly panicking.

Trigger: First major SaaS company reports revenue miss (not just margin miss) — the move from Step 1 (multiple compression) to Step 2 (earnings miss) in Adam Parker's sequence.

Names: Avoid CRM, ADBE; underweight any SaaS-heavy index or private credit fund.

Investment Thesis #3: Physical AI Shortage Is Inflationary — Own the Atoms

Argument: After 17 years of "bits over atoms" (smartphone era), the AI buildout requires massive re-investment in physical infrastructure: gas turbines, copper wiring, silver for electronics and solar, DRAM for inference, and chips. The US has underinvested in all of these. Jensen Huang's three requirements for AI — power, chips, talent — are both scarce and inflationary. Visser's CPI forecast: north of 4% by mid-2026, driven by diesel, plastics, fertilizer, and energy costs.

"We don't have enough copper. We don't have enough silver. We don't have enough energy in general. We certainly don't have enough electricity. We need gas turbines. We need all of these things that we don't have."

The market is calling commodity inflation "transitory" — just as it called COVID inflation transitory in 2021. It is not transitory; it is structural underinvestment from a decade of software-driven markets ignoring atoms.

Trigger: CPI print north of 4%. Subsequent prints show diesel and plastics costs filtering through. Silver breaks above its $78 consolidation toward prior high of $120.

Names: Silver miners (SLV, PSLV, AG), GE Vernova (GEV), Freeport-McMoRan (FCX), Albemarle (ALB), Lithium Americas (LAC), Brazil (VALE, EWZ).

Investment Thesis #4: Scarcity Portfolio — DRAM, Nvidia, and the AI Stack

Argument: Visser's personal portfolio is 100% oriented to "the east-west movement of Blackwell and Vera Rubin" — every component the agentic AI wave requires at physical scale. The highest-conviction positions: Micron (was $100 on his public list, now $400+), Nvidia (bought after Jensen's Dwarkesh interview; "too cheap" at decade-low PE despite 60–70% earnings growth), Cadence and Synopsys (Jensen explicitly named both; both down YTD = buy signal), Marvell Technology (agentic networking and custom ASIC for hyperscalers), Pure Storage (flash storage for AI workloads). Qualcomm added as a 6–9 month ahead edge-device play.

"Just download all of Jensen Yuan's transcripts for the year. Put them into one knowledge brain. Have Claude go in there and just ask it what companies to buy. He basically is a walking insider information of what to buy."

Nvidia has gone sideways for 10 months despite 60–70% earnings growth — the cheapest PE in a decade. Multiple compression has occurred; any DeepSeek-triggered sell-off is a buying event, not a thesis-breaker.

Trigger: Nvidia forward PE re-rates toward Caterpillar's 36×. Cadence/Synopsys break out above year-to-date resistance. Micron HBM ASP revision higher.

Names: NVDA, MU, CDNS, SNPS, MRVL (Marvell), PSTG (Pure Storage), QCOM (edge, 6-9 months ahead).

The Ecosystem Map

Core positions: Bitcoin (BTC), Nvidia (NVDA), Micron (MU), silver miners, Cadence (CDNS), Synopsys (SNPS), Marvell (MRVL), Pure Storage (PSTG), Albemarle (ALB), Lithium Americas (LAC), Brazil mineral exposure (VALE/EWZ)
AI tools used daily: Claude (primary for code and research), OpenClaw (local Claude deployment), Gemini (fact-checking, visuals), Grok, ChatGPT, Perplexity — "all of the above"
Ethereum and Solana: Owned but framed as temporary. "I believe they're going to outperform during the network effect days over Bitcoin" — but no durable moat; will ride altcoin season then exit
MicroStrategy (MSTR): Largest public Bitcoin proxy; Saylor as "the single largest owner" is an important structural buyer; adds floor support
Brazil / emerging market miners: Named as mineral producers (copper, silver feedstock) benefiting from physical AI buildout

Key Risks

Deep Seek V4 built only on Huawei Ascend chips: Would validate China's chip independence thesis → immediate semiconductor sell-off. Visser's response: buy the dip. But this is the single biggest near-term binary for semis.
Inflation stays transitory: If CPI does not breach 4% or 3-month T-bill yields, the Bitcoin regime trigger does not activate. Thesis takes longer.
Jevons paradox saves SaaS: If AI actually increases software demand rather than destroys it (as happened with electricity and cloud), terminal value compression reverses. Visser dismisses this but acknowledges it as a distribution-of-outcomes tail.
Private credit contagion: SaaS terminal value destruction hits private credit portfolios (25–50% SaaS exposure) → credit tightening → macro risk-off event could take down Bitcoin alongside everything else in the short term.
Fed raises rates: If labor market surprises strong and CPI forces the new Fed chair's hand, negative real yields do not materialize → Bitcoin regime delayed.
Quantum risk overhang (for Bitcoin): Still the #1 objection from traditional finance; Visser believes it is mythical but acknowledges it suppresses Bitcoin adoption.

Investment Opportunities at a Glance

Tier	Name / Category	Core Thesis	Conviction Signal
1	Bitcoin (BTC)	Last surviving moat; AI destroys all code-based stores of value; scarcity is the only defense	"247% annualized" in the negative real-rate regime; 100% of BTC returns in this quadrant
1	NVIDIA (NVDA)	Decade-low PE; 60–70% EPS growth; any DeepSeek sell-off is a buy event; "play from strength"	Bought right after Jensen Dwarkesh interview; plans to buy more on any sell-off
1	Micron (MU)	Was $100 when pitched, now $400+; DRAM/HBM structurally scarce; the scarcity-side memory play	"I kept pitching Micron when it was $100 and sub-$100" — multi-year validation
2	Silver miners (AG, PSLV)	Needed for semiconductors, solar, drones; finite supply; up from $60 low, secular bull	"Unlimited upside because there's a limited supply" — still below $120 prior high
2	Cadence Design Systems (CDNS)	Jensen-named in Dwarkesh interview; down YTD = opportunity; EDA tools multiply with AI agentic chip design	"When Jensen says two names on there and they're down for the year, Jordi's going to buy them"
2	Synopsys (SNPS)	Same Jensen-endorsed thesis as CDNS; down YTD; agentic chip design multiplies seat count	Explicitly named alongside Cadence as near-term buy
2	MicroStrategy (MSTR)	Largest public Bitcoin proxy; Saylor perpetually buying; structural floor support	"He is the single largest owner and always will be"
3	Marvell Technology (MRVL)	Custom AI ASICs for hyperscalers; agentic infrastructure; Jensen-ecosystem play	Named in personal portfolio as part of the Blackwell/Vera Rubin buildout basket
3	Pure Storage (PSTG)	Flash storage for AI inference workloads; named in personal scarcity basket	Explicitly named alongside Micron and Marvell in portfolio disclosure
3	Albemarle (ALB) + Lithium Americas (LAC)	Lithium for batteries (humanoids + EVs); physical scarcity; reshoring tailwind	Named in personal portfolio under "lithium for batteries"
3	Qualcomm (QCOM)	Edge AI devices; down 20% YTD; 6–9 months ahead of when market will price edge story	"I just did one on edge devices... the companies that will benefit like Qualcomm"
3	Brazil / VALE (EWZ)	Mineral producer (copper, silver, nickel); physical AI buildout feedstock	"Brazil because Brazil is a mineral producer" — named directly in portfolio
4	Ethereum (ETH) / Solana (SOL)	Temporary network effect play; traditional finance needs DCF-able assets → ETH/SOL for altcoin season	"When the ETH vs BTC chart breaks out, I will probably go buy a bunch of altcoins... but not for very long"

Monitoring Checklist

Bottom Line

The AI trade is not the S&P 500. Visser argues the S&P will be flat in nominal terms over the next 10 years while the economy doubles — productivity flows to entrepreneurs and physical asset owners, not to existing corporations. The only reason the index is up in 2026 is semiconductors and AI infra; the other 80% is suffering.
Bitcoin's activation regime is approaching. 100% of Bitcoin's historical returns have come when CPI exceeds 3-month T-bill yields with the Fed on hold. That quadrant is weeks away per Visser's inflation forecast. This is not speculation — it is a historical regime switch backed by 15 years of data.
The DeepSeek-on-Huawei scenario is the clearest trade in the market. If Deep Seek V4 achieves Opus 4.6-level performance running only on Huawei Ascend chips, Nvidia sells off sharply. Visser's explicit plan: buy more. A sell-off in NVDA, MU, CDNS, or SNPS on a China AI headline is the entry, not the exit.
Silver is the most under-owned scarcity play. Every semiconductor, solar panel, drone, and humanoid robot requires silver. Supply is finite. The metal has base-built from $60, is at $78, and has a prior high of $120 as the next level. No AI portfolio Visser knows of owns it.