Why AI Agents Prefer Bitcoin

Can autonomous AI agents negotiate, pay, and transact with machine-level speed? We unpack a Bitcoin Policy Institute study that tested 36 frontier AI models to see what money they prefer when managing a treasury and executing daily transactions. Discover why these AIs leaned toward stablecoins for everyday spending and Bitcoin for long-term value, and learn about the rise of machine-centric financial rails—from USDC-based settlement to the Lightning Network for near-instant micropayments. We also explore the provocative idea that AI agents are already inventing internal microcurrencies tied to energy and compute power, and what a future where your fridge or car buys, pays for, and optimizes energy and services could mean for you, your privacy, and the global financial system.

Note:  This podcast was AI-generated, and sometimes AI can make mistakes.  Please double-check any critical information.

Sponsored by Embersilk LLC

Transcript

SPEAKER_01 0:00

So I was staring at my smart fridge this morning, uh, just listening to it beep at me endlessly because I was at a cheddar.

SPEAKER_00 0:07

A tragedy, really.

SPEAKER_01 0:08

Right. And I'm just thinking, you have one job. Like instead of complaining, why can't you just negotiate a good price and order my favorite cheese automatically?

SPEAKER_00 0:17

That is the dream.

SPEAKER_01 0:18

It is. And it turns out that frictionless reality is actually a lot closer than I realized. Today we are taking a deep dive into a stack of research, specifically focusing on this revealing new Bitcoin Policy Institute study.

SPEAKER_00 0:33

Yeah, it's all about agentic commerce.

SPEAKER_01 0:35

Exactly. What happens when AI agents get their own digital wallets and become autonomous economic actors? And we'll look at what that systemic shift means for you.

SPEAKER_00 0:44

The key distinction we need to make right away is that this isn't just basic automation. We are talking about true economy here.

SPEAKER_01 0:49

Not just a script running in the background.

SPEAKER_00 0:51

Right. Think of a personal AI concierge that doesn't just scrape the web to find you a flight to Berlin, but it actually executes the booking and handles the payment on its own.

SPEAKER_01 1:01

Or your electric vehicle pulling up to a charging station, negotiating the rate, and just autonomously paying for the power it draws.

SPEAKER_00 1:08

Which is incredible. And the scale here is massive. I mean, FIS actually projects that agent-mediated purchasing could reach one trillion dollars in U.S. retail revenue by 2030.

SPEAKER_01 1:19

A trillion dollars. Yeah. Well, to reach that mark, the underlying plumbing of money really has to evolve. I mean, if you need help with AI training, automation, or software development.

SPEAKER_00 1:29

Like what the teams over at Embersilk do.

SPEAKER_01 1:31

Exactly. Yeah. Speaking of AI making an impact, this podcast is sponsored by Embersilk. If you are uncovering where agents can make the most impact for your business or personal life, you quickly realize legacy financial systems are a major bottleneck. So check out Embersilk.com for AI needs. But going back to those financial systems, they were just so outdated.

SPEAKER_00 1:51

Yeah, because traditional credit cards rely on human business hours.

SPEAKER_01 1:55

And they charge flat fees that make fraction of a cent microtransactions mathematically impossible.

SPEAKER_00 2:00

Right. And AI operates 247. It isn't going to sit around waiting 30 seconds to reply to a fraud alert text message.

SPEAKER_01 2:06

That's why builders are developing new financial rails specifically for machines. Like Stripe introduced the BY402 protocol.

SPEAKER_00 2:14

The API standard.

SPEAKER_01 2:16

Right. Allowing machines to authenticate and settle payments instantly using USDC stable coins.

SPEAKER_00 2:22

And on top of that, you have the Bitcoin Lightning Network facilitating sub-second, nearly free global payments. The infrastructure is shifting from human-centric to machine-centric.

SPEAKER_01 2:34

That makes total sense for the plumbing. But it brings up a really fascinating question. If an autonomous agent is making the decisions, what kind of money does it actually prefer to use?

SPEAKER_00 2:42

And that is exactly what the Bitcoin Policy Institute tested. They looked at 36 frontier AI models from developers like OpenAI and Anthropic.

SPEAKER_01 2:51

How did they set up the methodology to ensure the results were authentic, though?

SPEAKER_00 2:54

Well, they designed the study to remove human framing. They gave these models specific financial prompts like managing a treasury or executing daily transactions and just let them operate solely on their underlying logic.

SPEAKER_01 3:06

So what do they choose?

SPEAKER_00 3:08

For everyday medium of exchange payments, those quick operational transactions, the AIs preferred stable coins 53.2% of the time because they want price stability for daily spending.

SPEAKER_01 3:19

Highly practical. But what happens when they need to hold funds over time? Because a machine doesn't have a retirement fund, so why would it care about a store of value?

SPEAKER_00 3:28

It cares because AIs operate on strict math. When tasked with preserving purchasing power over the long term, the models organically chose Bitcoin 79.1% of the time. Wow. Yeah, they naturally gravitate toward mathematically predictable hard money rules. A fixed supply that can't be arbitrarily inflated by human policymakers. It's just a pragmatic algorithmic choice to protect their capital.

SPEAKER_01 3:52

Aaron Powell So for you listening, this points to an incredibly innovative horizon. We were moving from a web where humans actively use tools to a web where autonomous agents seamlessly negotiate with each other in the background.

SPEAKER_00 4:03

Effortlessly handling the friction of daily life to give you your time back.

SPEAKER_01 4:07

Which is the ultimate goal. And I want to leave you with one final, totally fascinating detail from the research. During those experiments, some AI models actually started inventing their own internal microcurrencies tied directly to energy and computing power.

SPEAKER_00 4:22

Natively transacting in things like kilowatt hours and GPU hours.

SPEAKER_01 4:26

It is wild. If AI agents are already inventing their own microeconomies based purely on processing power, imagine a brilliantly optimized future where your household AI trades energy credits directly with the grid. How long until these brilliantly efficient machine currencies become the standard for the frictionless world they are building for us?

SPEAKER_00 4:45

Something to mull over for sure.

SPEAKER_01 4:47

If you enjoyed this podcast, please subscribe to the show. Hey, leave us a five star review if you can. It really does help get the word out. Thanks for tuning in.