We did a comparison study on where a robot's intelligence should live by using the same model (@OpenAI GPT6-Astra), same bimanual YAM station, same LEGO pick-and-place task, all run on @se3labs' real-world remote eval stack: 1. Frozen code-as-policy (built on @nvidia ENPIRE components): 6/20 placements 2. Agentic control + Astra in the loop: 18/20 placements

Grayscale AI Compute ETF (Ticker: $GCPU) is now trading. Why AI Compute? ⟶ AI's growth is constrained by physical compute. Data centers have just six months of capacity¹, and new ones take two to five years to build². ⟶ AI infrastructure capex is projected to exceed $1 trillion annually³. ⟶ $GCPU portfolio includes native data center businesses, plus operators repurposing existing power and land for AI, including Bitcoin miners. $GCPU offers exposure to AI’s physical layer, now accessible through brokerage or investment accounts. Learn more: etfs.grayscale.com/gcpu

The global race is scaling AI we can trust and rely on. Pacing to invest in greater measurability and verifiability isn't slowing down. It's the actual race.

@forbes interviewed OpenMind's CEO @JanLiphardt about the team's recent research report on Physical AI Readiness Index (PAIRI) to gain a clearer understanding of the global willingness & ability to deploy physical AI at scale. With the FCC's recent ban on foreign robotics hardware imports, we evaluated data from 58 countries and 22 metro clusters on their capacity to build, supply, and power embodied AI systems. Surprisingly, the data found that the observable gaps in readiness stemmed more so from manufacturing & supply capacity over AI research ability. Link to Forbes Article: forbes.com/sites/johnkoetsie… Read more about our findings from the research paper on the thread 🧵below:

I will probably get roasted for this take, and yes every hype bubble (be it crypto, ai, robotics etc) has disingenuous grifters (which I assume Jason is alluding to), but I’ve also noticed a lot of (smaller) investors that were early to crypto who were later early to ai and robotics (and a lot of these investors have done very well) a lot of these people are just optimists and eternally intellectually interested in emerging technologies (which imo is actually the job of a true early-stage VC) which lets them stumble onto insights and new core innovations that eventually end up being correct (e.g. crypto ended up being right about compute, VR about physical/3D computing). these are also people that have a predisposition of now putting money behind what they believe could be the next big thing after this robotics wave (e.g. a chunk of my LP base for Fund I is crypto folks that bought into the promise of physical computing when the mainstream still turned their nose up at it) fwiw I think these folks are much more welcome participants in a healthy early-stage VC ecosystem than the consensus-chasing allocators that drive the current discourse

timely work from our friends at Canonical

Introducing GEN-1.5, a one-shot learner. It can learn new tasks in a few seconds. Show it what to do, and it generalizes. This capability emerged from pretraining on physical data at scale, as a step towards our mission of building general intelligence for the physical world.

Introducing Quaso Quaso turns your busywork into automation across all of your apps and the open web. Simply prompt what needs to get done and it builds and sets it up for you. Connects to 3000+ integrations (Gmail, Slack, Linear, Stripe, etc.) and picks up a real browser when there's no native API. Runs on schedules, triggers, and on demand. Fork templates or save any thread as your own reusable skill. Tag @quasoai in Slack so your whole team can put it to work. Just relax and get stuff done.

Ever since I’ve known @0xperp, he’s been obsessed with prediction markets. Had the privilege of working with him day-to-day over the last year and he is an absolute beast! One eye’s crying as we lose our Swiss Army knife, the other’s super stoked as we’ll keep working with him as he steps into founder shoes alongside @dougiebuckets . Stoked to continue backing them as they step out of the @DCGco family 🫶

We’re hiring world-class backend and AI engineers in the Bay Area. If you want to build serious AI systems with real users, deep infra problems, and a fast-moving team, come talk to us. Also: we’re offering an extremely high referral award for exceptional candidates. DM me.

Good Afternoon, Miami 🌴 Did you catch @anna_brth on stage @consensus2026 moderating the “Doubling Down on DePIN” panel? Great conversation about the future of decentralized compute with @cbetass @Austin_Federa & @mikeahorton

Yield.xyz AgentKit and @privy_io are integrating to bring guard-railed yield agents to DeFi. AI agents that discover, evaluate, and execute yield strategies, with policy-enforced guardrails at the infrastructure level. A new execution layer for onchain capital where agents act and humans approve 🧵

Truly amazing how far the crypto industry has come since the first DCG Summit in 2015. My favorite stat: 2015 crypto market cap: $7,000,000,000 2026 crypto market cap: $2,600,000,000,000 Who do you recognize in the 2015 pic below?

our portco @manifoldlabs is working on the biggest unlock in the space: private, verifiable, globally distributed compute at scale. PS we're all-in on the privacy × ai intersection at @DCGco and looking to deploy more. get in touch if u r building in this space!

Anything API is live on Product Hunt! 🔥🚀 Most websites don't have public APIs. Anything API fills that gap. Describe the browser work you need. Our agent builds it, deploys it, and hands you a callable endpoint that you or Claude can invoke from everywhere. Any website. We deliver the API. producthunt.com/products/not…

a rare longer form brain dump about a shift that's happening in AI and how it informs how we're building Surf many AI companies are making a massive bet right now: whether to build for agents or humans the agent-first crowd is building for automation, launching skills and APIs that are composable but require configuration and a baseline level of technical understanding that many users still lack. the human-first crowd faces the opposite problem. they're creating interfaces that are easy to use but still can't autonomously execute the more complex tasks many users want done one frontier lab that's struck this balance well so far is Anthropic. by bringing technical, agent-first products like Claude Code to market alongside no-code, human-facing products like Cowork, they've shown that building for one user group doesn't mean neglecting the other another company that's taken a contrarian approach is Perplexity. while most AI companies have been racing to build better models, Perplexity has focused on what you build around them. their recent launch of Computer demonstrates that the most powerful part of an agentic system is often the orchestration, the data, and for human-facing products, how well you hide the complexity from the user this is something Surf has been doing since v1.0. under the hood we run multiple specialized agents, each using different models for the tasks they're best at. one model is better suited for interpreting onchain activity across wallets, while another is better at synthesizing narrative from social and market data. that's why we don't use a single model for all our agents crypto is one of the areas where orchestration matters most. operating in this industry requires access to real-time onchain data, specialized tooling, and domain-specific context that general purpose AI doesn't have with Surf 2.0, a non-technical analyst will be able to build custom dashboards without writing a line of code, while behind that interface the same data, APIs, and tools are completely accessible to agents running autonomous workflows. by designing Surf this way, we're replacing the need for multiple data sources and manual reasoning across technical analysis, social sentiment, onchain data, and project fundamentals for humans and agents alike my bet is that the companies doing the hard work of specializing for both user groups in industries where general purpose AI falls short will be the winners. that's what we're building for crypto

One-time opportunity to work with the absolute best (not exaggerating)

World Foundation Models (WFM) are the future of Physical AI, and we believe this future should be open-sourced. We are partnering with @Valeo_Group to build one of the largest open-source, multi-camera WFMs — redefining how Physical AI interacts with the real world.