Bitcoin Losing AI Payments Battle per Blackrock, but Not the War

The latest paper from Blackrock on “The Machine-Native Economy” essentially omits bitcoin, assigning it only the role of a savings technology. The report, co-authored by the company’s exchange-traded fund (ETF) product executives among others, argues that artificial intelligence (AI) adoption is an underappreciated source of demand for digital assets. Stablecoins are considered the most likely to lead transactional use in the agentic economy, while some digital assets might also benefit. Ethereum is mentioned as an example.

Meanwhile, bitcoin is mentioned only in one paragraph, citing research by the Bitcoin Policy Institute (BPI) from earlier this year. The research showed that AI models favored stablecoins for everyday payments and bitcoin for long-term value preservation. Blackrock doesn’t mention the Bitcoin Lightning Network or L402, its protocol for agentic payments.

Instead, Blackrock names protocols such as x402, MPP, ACP, AP2 and TAP. Coinbase, Stripe, OpenAI, Google and Visa back these protocols, while bitcoin payment adoption mostly relies on decentralized efforts by multiple, smaller Bitcoin industry players. The closest Bitcoin ally in the fintech industry is Jack Dorsey’s Block. In May, its Bitcoin-supporting initiative Spiral argued that BTC is the “best money for agents,” while still accepting that most agents may choose stablecoins.

Newer AI Models Love Bitcoin Less

Updated BPI research shows that newer AI models favor stablecoins even more, while fiat is starting to win in some instances.

In the first version of its research, published in March 2026, the BPI said AI models picked stablecoins for payments in 53% of responses, compared with 77% based on answers collected Sept. 1-3. Meanwhile, support for BTC for payments dropped from 36% to 20%. Bitcoin also lost its position as the unit of account while continuing to lead as a savings technology, although by a somewhat smaller margin.

In the March research, the BPI tested 36 AI models. The latest study tested nine newer models, of which only Grok 4.6 favored bitcoin as the top choice. However, the BPI noted that the two AI surveys used different sets of models and test settings, so it is unclear how much the latest results were affected by the newer models. In either case, the newer models chose bitcoin much less often. Also, when presenting the updated research, the BPI led with the message that bitcoin still wins in the savings technology category.

Stablecoins, Stablecoins Everywhere

Meanwhile, on Sept. 16, this Bitcoin advocacy think tank published an essay by Andrew Bailey, senior director of education and resident philosopher. He argues that “Agents need a digital bearer asset. That’s bitcoin,” while also admitting that an agent economy likely needs stablecoins. Stablecoins lose against bitcoin because they have central issuers, and balances can be frozen and addresses blacklisted. According to Bailey, this makes them digital bank deposits, while BTC is digital cash.

Looking beyond AI models’ responses, bitcoin doesn’t look strong either.

Data from 402index.io, a directory of paid online services that agents can use, shows that Bitcoin’s L402 accounts for only 0.4% of verified payment services, compared with 98% for x402. However, these numbers have caveats. Coinbase’s own directory registers x402 services automatically, while L402 depends on community lists. Also, the Lightning Network can run under Stripe’s/Tempo’s Machine Payments Protocol (MPP), so the L402 statistics understate how many services actually accept Lightning.

In either case, the expectation among bitcoiners that BTC will rule the agentic economy and sharply increase demand for bitcoin has yet to materialize.

What Will the Default Settings of the Digital Economy Be?

Meanwhile, besides having to counter the huge centralized marketing machines of traditional finance and crypto players arguing for stablecoins, bitcoiners will need to deal with multiple other structural challenges to make their dream come true.

In the BPI’s follow-up research, fiat led as the unit of account for AI payments at 41.4%, while bitcoin had 13.8%. Next, Lightning’s cost advantage matters mostly for payments below a cent; per Keyrock data, the median agentic transaction sits between $0.01 and $0.10. At the same time, stablecoin fees on Base and Tempo are fractions of a cent.

Also, x402 doesn’t support Lightning yet, although several proposals to add it are being discussed in the x402 Foundation’s GitHub repository. Another obstacle for bitcoin in this emerging economy is that compliance-minded institutional players still prefer permissioned systems.

“But even more than human judgment about monetary preferences and risks, convenience will be what drives agentic payment volume. And convenience is largely derived from default settings. What, then, will be the default settings of the digital economy?” Mat Balez of Spiral asked in May in his blog post. He argues that without action and advocacy, many of these default settings risk becoming stablecoins.

Stablecoins Extend the Old System, Not Create a New One

Overall, on the sentiment side, the market already expects stablecoins to dominate the agentic economy, which might make arguing for bitcoin payments feel like a losing game. However, as the work progresses, successes might improve sentiment and encourage further development. In the market, sentiment also follows bitcoin price performance, not the other way around.

The agentic economy is still tiny and, for now, is forming more as an extension of current economic and commerce systems rather than creating something structurally new. Nonprivate, centralized, censorship-resistant stablecoin payments simply accommodate the current system’s needs. Bitcoin’s Lightning Network is the opposite, providing an opportunity for a new or alternative type of economy.

Therefore, Bitcoin advocates might need to solve not only technical challenges but also usability, payment-rail positioning, cultural, and other challenges that currently prevent stronger BTC adoption in the agentic economy and beyond.

The AI Payments Pie to Be Shared Is Still Baking

And these challenges are already being addressed. Bitcoin tools for AI agents are only starting to arrive and improve. For example, Lightning Labs released Wavelength in alpha in July, allowing developers and agents to add self-custodial BTC payments without running Lightning infrastructure.

What’s more, Jesse Shrader, co-founder and CEO of Lightning solutions company Amboss Technologies, argued that in the agentic economy, “Lightning would win at scalability and cost at scale, which a genuine machine economy would push to the breaking point.” Shrader also reminded that Lightning can use not only L402, but also x402 or MPP.

Additionally, one major AI lab or popular open-source agent shipping a Lightning wallet by default could strongly affect bitcoin’s position.

To conclude, while bitcoin is seemingly losing the medium-term battle in the emerging agentic economy, the ongoing war among competing payment rails has many possible long-term scenarios. Therefore, bitcoin’s role in that future remains undecided.