i spent monday afternoon meeting with the @SECgov Crypto Task Force in Washington DC the topic: AI, Crypto, and America joining me were four of my @collab_currency portfolio companies building in the open source and decentralized AI space @NousResearch (@dillonrolnick, COO) @PrimeIntellect (@jimmy_811, COO ) @404gen_ / @opentensor (@6329, CEO) @actualinc (@Tom_A_Lynch, CEO) the @SECgov Crypto Task Force came ready to learn we all left blown away by their curiosity, thoughtfulness, and their clear respect for teams working on hard problems across the AI and blockchain spaces over the course of 90 minutes, our discussion with the crypto task force ranged from the benefits of building decentralized AI protocols, the role of energy markets on SME, america's unused/idle compute and utilization rates, along with new innovations in post-training that favor open infrastructure (e.g. reinforcement learning) we also discussed the role of applying blockchains and crypto to open source AI systems, where unique markets can form around any job across the AI stack in distributed settings -- data, compute, training, among others the north star view i shared was simple: using blockchain-based networks, i believe it will be possible to build AI systems in the future with (a) SOTA performance, (b) greater transparency, and (c) more equitable ownership for millions of american citizens these milestones will not happen tomorrow but they will happen and they'll be accomplished by teams like @nousresearch, @primeintellect, @404gen_ / @opentensor, @actualcomputer, and hundreds of other american projects building with a bias toward open and community-owned infrastructure thank you to @collab_currency COO @zackrosenblatt_ for helping organize the killer event thank you to @SECgov Crypto Task Force Chief Counsel @michaelselig, and to the other members of the Crypto Task Force for your generous time we were grateful to share a few of these ideas with the task force and deeply proud of the work you're putting in on behalf of the country

Introducing… Raster! A new digital art marketplace from @0xmetaclass and @thefunnyguysNFT. We’re making art collecting delightful by solving digital art's biggest headache: fragmentation.

🍜

@Dorm_DAO’s second annual “nyc interns & alumni dinner” — this year w/ sushi making 🍣 few things are more fulfilling than watching our brilliant alumni thrive in industry return to inspire the next class of leaders. @Dorm_DAO is family ❤️

genesis by @clairesilver created april 2022 early expressions of ai art, etched on-chain

hyperstrategy

Releasing INTELLECT-2: We’re open-sourcing the first 32B parameter model trained via globally distributed reinforcement learning: • Detailed Technical Report • INTELLECT-2 model checkpoint primeintellect.ai/blog/intel…

Reinforcement Learning in the era of LLMs requires scalable, distributed systems to push the boundaries of reasoning and alignment. Today - we release Atropos - our RL environments framework. github.com/NousResearch/Atro… Atropos is a rollout framework for reinforcement learning with foundation models that supports complex and diverse environments for advancing the capabilities of foundation models. In Greek mythology, Atropos was the eldest of the three Fates. While her sisters spun and measured the threads of mortal lives, Atropos alone held the shears that would cut these threads, determining the final destiny of each soul. Just as Atropos guided souls to their ultimate fate, this system guides language models toward their optimal potential through reinforcement learning. The work on Atropos was led by @dmayhem93 and built alongside @teknium, @rogershijin, @max_paperclips, @nullvaluetensor, @JSupa15, @artemsya and @karan4d

We did it — the first decentralized RL training of a 32B model is complete! Full open-source release is coming in ~1 week, including: checkpoints, data and a detailed technical report.

I spoke with @awrigh01, co-founder and CEO @tributelabsxyz for the latest episode of the Slow Hunch pod. We explored the evolution of organizational structures from joint stock companies to DAOs, and how AI agents are increasingly becoming part of these decentralized systems. At @usv, we've backed Aaron's vision of thin technology layers enabling open coordination and collaboration at scale - from hundreds to potentially millions of people, "just like the internet wanted it" in Aaron's words. :)

1/ Dorm Builders 🤝 Endgame Summit After 15 weeks of building, we’re excited to have our eight student teams showcasing their Bittensor projects on stage at @Endgame_Summit in Austin! Shoutout @collab_currency, @opentensor, @yumagroup, @fish_datura, @mogmachine &  @JosephJacks_ for providing this unforgettable experience to our squad❤️

AI x Crypto last month at my investor AGM in new york i gave a < 8 minute lightning talk on the AI x Crypto intersection briefly covered both infrastructure and application layers check out the talk if you’re interested, i created it to be approachable for all audiences i'll also add here a few policy points that didn’t make my talk across the last few hundred years, america has created unprecedented opportunity for individuals and businesses, in large part because of our country’s long revered bias towards market competition and entrepreneurship on the eve of AGI, i believe we have a responsibility to renew our identity as a leader on these dimensions i think this can be accomplished, in part, by orienting our country’s AI policy goals towards crypto based systems i dive further into this idea in my essay from january: x.com/derekedws/status/18762… empowering these two technologies to integrate together will likely catalyze widespread participation in open source and community-owned networks, through incentivized contribution at all layers of the AI stack over time, this will stimulate richer competition in all categories of AI creation, and encourage new forms of bottoms-up mobilization and broader social impact and best of all blockchains allow any individual to provably own the digital objects, work, systems, and infrastructure we’re collectively creating around AI and synthetic intelligence systems anyway no one has all the answers here, but i continue to believe there’s a north star for humans at this intersection i hope you enjoy the talk and ideas presented

dTAO subnets went live on feb 15 last two months have already been a blast decentralized intelligence markets over the internet believe in something (in real time) running bittensor (h/t @ronangmi)

Bitcoin & Ethereum were open‑source battering rams—smashing legacy finance and unleashing real competition. It’s time to reload and take on capital markets and media.

Next Thursday, we'll be welcoming 60 student leaders to @solana's NYC HQ for @Dorm_DAO's second annual summit, presented by @artemis. We started @Dorm_DAO's second year by welcoming 8 new clubs to the DAO family. Now, we celebrate a year of investing, building & catalyzing. 🗽 🍜 = 🙌

As part of the launch of Unsupervised Capital, we’re excited to announce our first Bittensor subnet token investment: @404gen_ 404-GEN is building the infrastructure layer for AI-generated 3D content. We believe it can become the most performant compute network for 3D asset generation—serving everyone from indie creators to AAA studios building AI-native games and virtual worlds. Today, 3D content creation is still mostly confined to gaming—a $16B market on its own. But the digital world is quickly reorienting around spatial, interactive formats. Social apps, productivity tools, and entertainment platforms are being rebuilt with 3D at the core—and demand for scalable, high-quality assets is exploding. Successfully operating a Bittensor subnet requires deep technical expertise, a strong community, and sharp product vision. Ben James (@6329) and the team at 404-GEN check all of these boxes. We couldn’t be more excited to support them as they push the 3D AI frontier forward and turn 404-GEN into the default engine for 3D content creation. Our investment memo is linked in the next tweet

the future is now

Today we’re launching INTELLECT-2: The first decentralized 32B-parameter RL training run open to join for anyone with compute — fully permissionless. Scaling towards frontier reasoning across coding, math and science.

⚙️ 404—GEN has released the WORLD'S LARGEST open-source 3D model dataset ⚙️ • 21.5M AI-generated 3d models • Open-source and attribution-ready • Larger than all existing 3D datasets combined This is an unmatched scale for Gaussian Splatting research. More details below 🧵 ⬇️