A brand new installment of Chain of Thought, the Brownstone Analysis newsletter written by Ben Lilly, argues that the battle over open-source artificial intelligence is following the identical path Bitcoin walked a decade in the past, and that traders who acknowledge the sample stand to revenue.
The observe opens with testimony that Anthropic CEO Dario Amodei gave to Congress in July 2023. Amodei acknowledged that open supply is “an excellent factor” in most scientific fields and that the dangers of open fashions launched to date have been “comparatively restricted,” however he warned that the scaling of open-source fashions was heading “down a really harmful path.”
Lilly reads the subtext plainly: if open fashions are harmful, then the closed fashions bought by firms like Anthropic are the protected selection — and the coverage that follows is to limit the open and elevate the closed.
Bitcoin’s early skeptics mirror what AI is dealing with
That framing is one digital-asset traders know effectively.
He revisits Bitcoin’s early skeptics, from Rep. Jared Polis shopping for the primary Bitcoin on Capitol Hill in 2014 to Sen. Joe Manchin’s name to ban a “harmful forex,” by the 2023 accusations that regulators tried to chop crypto off from the banking system in what critics dubbed “Operation Choke Level 2.0.”
The business survived, he notes, and Washington is now shifting towards clearer guidelines by the passed GENIUS Act and the pending CLARITY Act.
Decentralized AI, which Lilly calls “DeAI,” is having that very same combat now. He factors to latest developments as proof the partitions are going up: a U.S. export ban on Anthropic’s newest launch, which he says will push the corporate towards permissioned entry that verifies a person’s identification earlier than granting a mannequin, and OpenAI’s choice to limit its GPT-5.6 rollout to trusted companions.
He expects identification necessities to unfold. “It’s in your safety, you see,” he writes. “It at all times is.”
The observe leans on a national-security anecdote to clarify the worry driving these strikes. Lilly cites NSA chief Joshua Rudd, by means of Sen. Mark Warner, describing how Anthropic’s “Mythos” mannequin broke into “nearly all of our labeled system, not in weeks, however in hours.”
But open supply is closing the hole, in line with the piece. Lilly says the latest GLM-5.2 scored on par with Anthropic’s Sonnet 4.6 from February, leaving open fashions roughly three to 4 months behind the frontier, and predicts an open rival to Mythos and GPT-5.6 by fall.
He argues the larger unlock is decentralized coaching on peer-to-peer networks that mirror Bitcoin and Ethereum — swapping compute-for-network-security for compute-for-model-training. Distributed coaching, he notes, has grown from sub-1-billion parameters to 100 billion in two years.
He names three early initiatives — Darkish Bloom, which allows low-cost non-public inference on idle Macs; c0mpute, a decentralized inference community; and Pluralis, which trains AI throughout distributed shopper GPUs — and expects extra to launch tokens and reward customers for contributing compute.
The observe ends with the notion that governments will attempt to ban open fashions and they’re going to fail. For him, investing within the area “might be like shopping for Bitcoin in 2014, again when it was nonetheless ‘harmful.’”
