gm Bankless Nation, Welcome to Mindshare by Bankless, our newsletter for all things AI x Crypto! This week, Nvidia unveiled GR00T, an open-source AI for humanoid robots, sparking buzz around decentralized robotics…
Today's Issue ⬇️
📈 Market Pulse: Bank(r) on Base With continued flows, Base szn is blooming.
📸 Market Snapshot: Riding high off last week’s surge, Base season continues to be in full bloom, netting a 20% jump in TVL this week amid onchain excitement and the launch of Coinbase's Verified Pools.
Back to $BNKR though. This position in mindshare comes as Caitlyn Jenner (yes, her) declared that “Base took one day to win [her] heart, sorry Solana,” before instructing Bankr bot to buy her some $DRB, causing the token to climb 200% on the day.
Quick refresher: $DRB is a memecoin born from “tricking” Grok into naming and launching a token via Bankr.
Outside of BNKR on Base, the standout AI agents in the green this week are actually the Bittensor-connected betting bots AION500 and DKING, which posted ~50% and ~25% gains, respectively.
Token prices as of 1pm ET
24hr
7d
VIRTUALS $0.68
↗ 5.6%
↘ 0.5%
AI16Z $0.18
↗ 1.8%
↘ 6.1%
ARC $0.06
↗ 2.5%
↘ 9.0%
. . .
AI ROLLUP
China's Next Frontier Model
China’s AI ambitions continue to heat up.
Baidu’s newest model just outperformed GPT-4.5 at half the cost of DeepSeek R1, demonstrating major efficiency.
Can OpenAI hold its ground here, or has the global AI power balance shifted decisively to the East?
In this episode, David and Ejaaz break down this big new breakthrough, plus they dig into the AI crypto market's drawdown, Google’s Gemini 2.0 Flash upgrade, and how hyper-realistic AI avatars are blurring the line between the real and the synthetic.
The age of robotics kicked off Tuesday during Nvidia’s latest keynote presentation, and, at least for now, it’s open source.
The AI titan introduced GR00T N1 — the world's first open-source foundation model designed specifically for humanoid robots. Think of GR00T as a universal software brain behind robots, enabling them to perform everyday tasks like picking up objects, pouring drinks, or even flipping pancakes.
Unlike simpler robots built for repetitive factory tasks, GR00T is designed to help robots handle complex, unpredictable environments — basically, real life.
Why go open source? A few reasons. First, Nvidia wants developers worldwide to build on its tech, speeding up robotics innovation. By also releasing massive training data — from videos of daily tasks to virtual robot simulations — Nvidia helps robots learn real-world scenarios better.
GR00T also stands out by blending cheaper simulated data with hands-on experience, helping robots train better for real-world deployment, and addressing a tough problem for robotics around the slow and expensive process of data collection.
Tesla, for instance, paid $48/hr for participants to generate high-quality robotics training data, like guiding robotic arms or navigating real-world environments. By blending synthetic simulations with real-world experience, GR00T will lower these costs, making robotics development more scalable.
But beyond the endless quest for innovation, there’s a bigger play in their decision to open-source: driving demand for Nvidia’s hardware and services.
It’s two birds one stone: open-sourcing GR00T simulatenously expands the market reach of Nvidia’s products, while also pulling more developers into Nvidia’s ecosystem.
All this ties into Nvidia’s broader vision for “Physical AI,” a term coined earlier this year to describe robots that interact intelligently with the real world.
Yet the implications go beyond just Nvidia. Crypto, being inseparably intertwined with AI, has picked up on this trend, fueling fervor around Decentralized Physical AI (DePAI). Thinkrobots operating on blockchains and all the synergy that can come with this new open-source momentum.
DePAI: Can Crypto and Robots Mix?
DePAI can be broken down into seven core layers, from hardware and governance to spatial intelligence.
While a handful of teams like Frodobots, peaq, and xmaquinaDAO are building across this stack, I agree with Knower and Smac of Compound VC that those building for distributed training may end up being the most important.
Remember, a huge hurdle for robotics comes from the costs and speed of amassing the real-world data needed to learn how to navigate the messy, unpredictable settings of everyday life.
Crypto comes in handy here, providing both the system and incentives to conduct large-scale, distributed training and to collect these precious bits of everyday data. Tokens can be used for compensation, while blockchains keep data sharing secure, transparent, and verifiable.
Several teams are already exploring this approach:
🎥Mecka.AI: Gathers videos of daily tasks — cooking, cleaning, general chores —to give robots a more realistic view of human behavior.
🏊PrismaXai: Captures first-person scenarios — sports, niche hobbies, unique environments — that are often overlooked by standard data-collection methods.
📱OpenMind AGI: Uses smartphones to record typical day-to-day activities like grocery shopping, fueling robot models with fine-grained real-world insights.
Additionally, many existing DePIN protocols like Hivemapper, Geodnet, and Natix Network may supplement robots with additional data to boost their spatial intelligence, equipping them with the data needed to also navigate the real world, rather than just act in it.
Supporters believe these approaches can rapidly scale data collection, helping robots learn faster and become more adaptable. Yet skeptics question whether crypto rewards will provide enough sustained motivation to gather the high-quality, consistent data that advanced robots truly need.
Privacy, too, remains a major concern for people asked to record personal moments. But if these hurdles are addressed, distributed data-collection models could unlock a scalable and affordable pipeline of real-world insights — giving DePAI the critical advantage it needs to propel general-purpose robotics into everyday life.
The Road Ahead
However, DePAI also faces some real challenges. Most blockchains today aren’t built to handle the enormous amount of real-time data (here's your cue, MegaETH) that advanced robots require.
Imagine 1,000s of delivery drones all needing split-second decisions to navigate busy skies — currently, networks would buckle under that kind of load. For DePAI to truly take off, blockchains will need to get much, much faster.
Then there’s the issue of interoperability — making sure all these robots, AI programs, and blockchains actually speak the same language. Right now, different robots from different companies use different standards, making collaboration tricky.
Hopefully, GR00T will go some distance towards establishing an industry standard, but only time will tell. Without this, we risk ending up with a fragmented mess in the near term, rather than the smooth-running, robot-filled future we’re picturing.
Overall though, DePAI holds promise in the same way crypto holds promise for AI, providing not just cheaper data collection, but also the potential to incorporate privacy-preserving tools to this data collection, incentives for open-source creation, and broader ownership of robotics tech.
Ultimately, Nvidia’s choice to open source GR00T as a way to speed innovation and juice distribution holds potential to align quite well with crypto’s own permissionless foundations, allowing robotics companies to acquire the vast pools of real-world data they desperately need and developers to experiment with applying cryptoeconomics to robots.
Yet, whether this synergy between robotics, crypto incentives, and open-source infrastructure truly revolutionizes the way robots enter everyday life or stalls amid practical roadblocks — well, only time will tell.
The Fraxtal ecosystem is expanding at lightning speed—this month’s biggest highlight is IQAI.com, the newest Agent Tokenization platform from IQ and Frax. IQ is building autonomous, intelligent, tokenized agents launching on Fraxtal in Q1. Empower onchain agents with built-in wallets, tokenized ownership, and decentralized governance—all within a fast-growing Fraxtal ecosystem.
Building an AI assistant and want it to be crypto savvy? Consider exploring our new Bankless Onchain MCP tool.
This resource lets you connect large language models (LLMs) to real-time blockchain data, letting AI assistants check token balances, analyze smart contracts, and track transactions.
The context is that MCPs, or Model Context Protocols, are becoming a major trend in AI, allowing models to access external tools and services without retraining.
As such, the Bankless Onchain MCP works like a blockchain translator for AI, fetching onchain data and delivering structured, easy-to-understand answers.
For users, this means you can now ask an AI simple questions like “What’s my ETH balance?” or “What does this smart contract do?” and get real, up-to-date answers instantly.
For devs, it offers a streamlined way to integrate blockchain data into AI-powered apps without dealing with complex infrastructure.
The tool is currently in beta and available for Bankless Citizens to experiment with, so if you're a tinkerer around onchain AI, now’s a great time to dive in and test its capabilities!
Not financial or tax advice. Bankless content is strictly educational and is not investment advice or a solicitation to buy or sell any assets or to make any financial decisions. This newsletter is not tax advice. Talk to your accountant. Do your own research.
Disclosure. From time to time, we may add links in this newsletter to products we use. We may receive a commission if you make a purchase through one of these links. Additionally, the Bankless team holds crypto assets.
