PodcastAnthony Pompliano ShowJordi Visser

Bitcoin Is the Toll Collector of the AI Economy

The discussion examines how AI agents could reshape banking, settlement, commerce, and competition among technology platforms, while raising questions about model economics, compute access, and where AI applications can build lasting value. Speakers also consider crypto's role in the agent economy, the conditions that could support its growth, and market risks ranging from higher rates to possible government intervention during severe financial stress.

BitcoinArtificial IntelligenceAI InfrastructureCrypto MarketsInterest RatesGovernment Debt

Preview · 20 more points in the app

Read the Full Summary

Key Takeaways

The speaker argues that AI agents could pressure bank economics from two directions: agents searching for better deposit terms may narrow banks’ deposit-investment spread, while faster stablecoin and tokenized settlement could reduce the float banks and card networks can invest. The speaker frames possible multiple compression over the next year as an opinion, not a settled outcome.

The speaker sees agent-focused commerce as a potential opening for new services and a challenge to established platforms: Paul Graham’s view is that Amazon’s restrictions on bots could create room for an Amazon designed for agents, while Shopify is leaning into agent-enabled shopping. The speaker also interprets restrictions by Apple, Amazon, and Google on outside agents as defensive signals of possible disruption, not proof that disruption has already occurred.

The bullets describe two possible pressures on general-purpose AI models: the speaker says Grok prompting other providers to open their systems could signal commoditization, while another speaker argues that compute access could let companies keep the most capable models from public release and supply them to businesses. Both are presented as views about possible industry directions, not established outcomes.

The speaker disputes the idea that AI token demand is already near a bubble, saying coding agents only recently began driving adoption and that companies are shifting toward more efficient usage. The speaker also argues that model economics depend on response time and the probability of a correct answer—not just token cost—citing a two-minute Opus 5.5 response versus a 12-minute answer from a cheaper open-source model.

The speakers see potential advantage in specialized AI applications alongside general-purpose models: Anthropic’s announced biotech focus is cited as an application opportunity, while one speaker argues that law and finance firms could fine-tune open-source models on their own data, subject to cost, talent, and other uncertainties. Another speaker predicts AI could enable one-person law firms, potentially resulting in a billion such firms.

TECHNOLOGY

AI Agents And Banks — The speaker argues that banks benefit from paying little or no yield on deposits while investing them in short-term Treasuries. AI agents seeking better terms could pressure that spread and banks' future valuations. The speaker presents multiple compression over the next year as an opinion, not a settled outcome.

“One is the rise of AI agents which means consumers are going from let's say lazy individual small businesses have their money in banks that pay them zero yield…”—

Faster Settlement — The speaker argues that same-day settlement using stablecoins and tokenization could reduce the float and investment opportunity available to banks and card networks. Longer settlement times give someone more time to invest funds in transit.

“So when you have the users, the consumers no longer being humans but being digital agents and at the same time you have financial guardrails which are being in the…”—

Agent-Focused Commerce — The speaker cites Paul Graham's view that Amazon's blocking of some bots or agents could create an opening for an Amazon designed for agents. The speaker also says Shopify is leaning into agent-enabled shopping and that an agent-friendly bank could be valuable.

“And he came out and I thought it was probably one of the most interesting perspectives this week was the first time in decades that he believes someone could expose…”—

ECONOMY

Government Backstops — The speaker argues that governments may put money into the system during severe financial stress, citing intervention after Silicon Valley Bank and pandemic-era support totaling trillions. The speaker presents this as a possible response, not a guarantee that markets would avoid losses.

“But I also know that if that happens, the governments have financialized everything…”—

MARKETS

Stripe Growth — The speaker contrasts Stripe's 10,000 employees with Citibank's 200,000 and cites Stripe's roughly $160 billion valuation and 41% year-over-year revenue growth, according to Stripe. The figures illustrate the company's scale and reported growth.

“This weekend I will go through about 15 slides of Stripe. City Bank is a $200 billion market cap company with $200,000 people. Stripe has 10,000 people…”—

Crypto Value Chain — One speaker argues that crypto's value has accrued not only to protocols such as Bitcoin and Ethereum, but also to underlying infrastructure and applications. This challenges the idea that value necessarily concentrates at the protocol layer.

“The equivalent in the AI is like the fat model lab thesis where all the value acrews to the model lab…”—

Crypto Toll Collectors — One speaker favors crypto's digital-money guardrails and toll collectors over investing in any single AI agent, arguing that many competing agents—including in Europe and Asia—could emerge. The speaker says they want exposure to areas where demand will exceed supply.

“I only believe there's two things to focus your attention on the toll collectors that's crypto as far as I'm concern. I want the digital money guardrails, the toll collectors…”—

FORECASTS

Crypto's Next Move — The speaker predicts that next year will bring the biggest moves in crypto because of AI agents. Traditional-finance participation could accelerate crypto's advance, the speaker says, and identifies higher tokenization volumes, tokenized-asset size, and stablecoin volumes as needed developments.

“And I'm telling you, the amount of traditional people that continue to bang on my door to talk about tokenization, to talk about how to make money on this, they're…”—

Treasury Yields — The speaker says 10-year yields are very likely to be around their current level a year from now. Yields could be lower if the jobs market weakens noticeably, which the speaker does not expect over the next year.

“I believe a year from now, it is very likely that 10-year yields will be about the same level a year from now as they are today…”—

Intro

The hosts introduce the “bankpocalypse” as a repricing risk, arguing that AI agents and faster settlement could weaken banks’ deposit spreads and middleman advantages without making banks disappear.

The “bankpocalypse”

They argue that agents may disrupt Amazon and banks by seeking better terms and bypassing incumbent services, while nimble firms such as Stripe may be better positioned to build agent-friendly commerce and finance.

Can AI agents disrupt Amazon & the banks?

The speakers disagree on model commoditization: one emphasizes Claude’s quality and compute advantage, while the other sees specialized applications, customized models, and infrastructure as sources of durable value.

Are AI models becoming commoditized?

They discuss how personal agents might monetize through transaction fees or improve Facebook’s advertising, while questioning whether conventional attention-based ads will remain effective as agents handle consumer tasks.

How do personal agents make money?

The guest favors compute and crypto as scarce infrastructure and financial toll collectors, while the host debates whether powerful model labs or customized business models will capture lasting value.

Toll collectors: compute & crypto

The guest says Bitcoin’s consolidation calls for patience, compares its technical setup with Micron’s, and expects agent adoption and rising tokenization and stablecoin activity to strengthen crypto.

Bitcoin outlook & why patience pays

They argue that prominent Wall Street research on crypto reflects a shift from contrarian thesis to broader acceptance, and that market returns depend on where future capital flows.

Crypto & Wall Street: contrarian to consensus

The guest rejects treating every warning as decisive bubble evidence, points to markets’ resilience through crises, and favors AI compute, crypto, and practical AI skills while acknowledging risks.

Bubble talk: signal vs noise

They emphasize distinguishing signal from noise amid rapid AI change, arguing that learning to use AI and tracking demand for compute can help investors and workers navigate uncertainty.

Jordi’s 2027 crypto world tour

Jordi jokes about a 2027 conference tour as he discusses sharing his crypto views more widely, then the hosts exchange event stories and close by talking about the guest’s book.

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal