Lord Rama crossed the ocean by building a bridge stone by stone, with an army of contributors, each doing what they were capable of. Dirgha is following the same model: we're bringing together people and organizations to build fully indigenous, agentic, sovereign AI computers.

A good thing that came out of the Anthropic restriction is my first video (2nd if you count the AI short). 98 more to go before I call myself a... vlogger? Is that term still in vogue?

💯 "If all the value is accrued by only a few models, the political economy will simply not tolerate it. There is no societal permission for an AI future that hollows out entire industries."

The permissioned path does not arrive as tyranny. It arrives as convenience. A society can lose its freedom this way without a single dramatic moment, simply by routing more of its thinking through infrastructure that answers to someone else. We must protect open source and open source models

No one expects Indian IT heavyweights to build frontier models. Isn't that telling? Freedom is earned by blood, sweat, and tears. You pay the price. Whether for a nation, a company, or a builder, true value demands a real cost. Technology concentrates power at an unprecedented scale, and advanced AI is accelerating this. Whoever controls frontier models and their underlying infrastructure gains massive economic, strategic, and scientific leverage. With that power, you have two choices. You can pioneer new science and build generational foundations. Or, you can use your capital and influence to protect short-term margins while remaining permanently dependent on others. The second path is what I call the chindi empire. "Chindi" is the Indian term for being stingy, miserly, and unwilling to spend where it matters. A chindi empire looks formidable on paper—hundreds of billions in revenue, thousands of talented employees, consistent profits—but it operates with a fearful, short-sighted mindset. It hoards cash. It avoids bold bets. It chooses the safety of services over the sovereignty of proprietary technology. Indian IT giants built global-scale companies from humble beginnings, creating millions of jobs and bringing in massive foreign exchange. But they took the path of least resistance. They had the capital and the talent to fund serious AI research and development years ago. Instead, they optimized for servicing other people’s tech. Now, the bill is coming due. Recent U.S. restrictions on access to advanced models like Anthropic’s prove that depending on foreign infrastructure is no longer a low-risk strategy. What is convenient today will be restricted tomorrow. Revenue from the services game doesn’t automatically buy technological capability. This is why true technological sovereignty matters. It isn't granted through access, and it isn't bought with safe bets. It requires allocating capital for the long term, embracing uncertain outcomes, and having the courage to build what you need instead of renting it. Every generation that built something meaningful paid for it. The architects of modern India didn’t play it safe. They took risks. They accepted that short-term comfort costs long-term strength. We are at the same crossroads with artificial intelligence. The chindi empire is tempting because it feels responsible. You protect margins, avoid spectacular failures, and placate shareholders. But over time, it leaves you vulnerable, dependent, and fundamentally hollow. I don’t want that future for Indian tech. I want us to build things that are our civilizational contribution to the world. Things that make history, push humanity forward. I want Indian builders to have real leverage in the AI era, not just implementation skills. This isn’t about whether TCS or Infosys should suddenly pivot into AI research labs. It’s about our ecosystem realizing that technological independence won't magically appear while we maintain the status quo. Freedom demands a price. Jai Hind. 🇮🇳🙏🏻

China is rapidly expanding its battery storage industry, betting that the future of clean energy belongs not only to those who generate power, but also to those who can store it for homes, factories and entire electricity grids.

56,000 tokens/sec at just 80 MHz. 🤯 I burned a full Transformer with KV cache into a custom chip. Designed gate by gate as a 100% digital integrated circuit. Prototyped on a FPGA. (No GPU. No CPU) Just pure digital silicon running @karpathy microGPT, spelling out names on a tiny LCD. This is GateGPT 👇

Carnegie Mellon’s Robotics Institute runs a course on robot learning... (For FREE 📌) 16-831 covers the full modern stack… the stuff actually being deployed right now: Imitation learning. Behavior cloning. Reinforcement learning. Learning from human videos. Sim-to-real transfer. Vision-Language-Action models. Not theory for its own sake. Every topic is anchored to a real robotics problem: how do you get a robot to generalize to environments it’s never seen before? All lecture slides are public. This is THE Robotics Institute. The place that produced the researchers now leading the frontier labs. Free. No login. 📌 [16-831-s24.github.io/lecture…] Follow for more robotics resources like this! —— Weekly robotics and AI insights. Subscribe free: 22astronauts.com

Top 10 Robotics Research Labs in India that need more attention and funding to build the next frontier Robotics and Physical AI models! Also for researchers, these are the top labs to aspire for! AIRL, IISc Bengaluru RRC, IIIT Hyderabad IITGN Robotics Lab Stochastic Robotics Lab, IISc HiRo Lab, CPS IISc MRL, IIT Kanpur Robotics Lab, Dept of Eng. Design, IITM FOCAS Lab, IISc Multi Agent Systems Lab, IISc CAIR, Bengaluru @ANRFIndia @NITIAayog @GoI_MeitY @IndiaDST

China just took AI infrastructure underwater. The country has reportedly launched a wind-powered underwater data center near Shanghai, designed to cool AI servers using seawater instead of traditional cooling systems 🌊🤖 Around 2,000 servers are said to be sealed inside submarine-grade modules, helping reduce land use, water consumption, and cooling energy. The project is powered by offshore wind and built for AI workloads, big data, and large-scale computing. Why it matters? AI is creating massive demand for electricity, cooling, and infrastructure. If underwater data centers prove reliable, they could become a new way to scale AI while using less energy. But the challenge is huge: corrosion, maintenance, sea conditions, and long-term reliability still need to be proven.

I'm building open-source, sovereign AI models, harnesses, and agentic marketplaces for the world. Dirgha Code, my open-source CLI, has recieved 18K organic downloads. Investors welcome. github.com/Dirgha-AI/dirgha-…

According to Grok, Andrej Karpathy is an EB-1 extraordinary ability green card recipient, not a US citizen. Thus under these new restrictions he is not permitted to use, or work on, Mythos 5 or Fable 5 as of 5:21pm tonight.

Ilya was right and predicted much of this

Things are not going well in the AI land. Today I woke up to this shocking news. It's funny, it's shocking. I don't know what to tell you. America just blocked access to its latest AI models for the rest of the world. For ALL of us. We have officially entered the haves-and-have-nots in the land of the free. Let me give you some context. In July 2025, this time last year, President Trump publicly floated the idea of denaturalizing Elon Musk. He is a naturalized American citizen. A trillionaire now. Think about that. A president suggesting he could strip citizenship from one of the most prominent businessmen, innovators, contributors in the country. Yesterday, that same playbook hit the AI world, the tech world. The Commerce Department ordered Anthropic to block access to its most advanced models — Fable 5 and Mythos 5 — for all foreign nationals. I mean all. And that explicitly included Anthropic's own foreign employees. Imagine that. The level of fear-mongering. The craziness that's happening. Unable to filter users selectively without breaking their infrastructure, Anthropic did the only thing they could: they disabled the models for everyone. So here's the pattern. Citizenship and national origin are now provisional gatekeepers. Not skill. Not contribution. Passport. More specifically, the birthright lottery. And it's always wrapped in the same language. Security. Fraud. Threats. Proliferation. That framing normalizes broad restrictions that function as tool of power. Control. What was once exceptional is now becoming routine. When a state can revoke belonging or restrict access based on where you were born, it doesn't create a two-tier system of capability. It creates a two-tier system of discrimination. Of permission. And permission—unlike code—does not scale. That's the world we're building. Let me know what you think in the comments.

Camera shy, nervous? Let AI take care of that and prepare you for real time.

India has <200 satellites in orbit. The United States has 10,000 . This gap is partially due to launch capacity. Skyroot isn't commercialised yet. But another limitation is satellite manufacturing. Which is why Hyderabad-based Dhruva Space plans to make two satellites DAILY!

👏🏻

Sir, they have blocked access to the very AI models you said we should share instead of build

We have officially entered the haves-and-have-nots era of AI. In July 2025, President Trump publicly floated "taking a look" at denaturalizing Elon Musk—a naturalized American citizen. Yesterday, the same pattern hit compute. The Commerce Department ordered Anthropic to block access to Fable 5 and Mythos 5 for all foreign nationals, explicitly including Anthropic's own foreign employees. Unable to filter selectively, Anthropic disabled the models for everyone. Citizenship and national origin are now provisional gatekeepers. Security framing—fraud, threats, proliferation—normalizes broad restrictions that function as control tools. What was once exceptional risks becoming routine. When the state can revoke belonging or restrict access based on passport origin, it creates a two-tier system not of capability, but of permission. And permission, unlike code, does not scale.

ETH Zurich just open-sourced their entire 2026 robot learning course. Not a MOOC. The actual course. Slides, lecture recordings, coding assignments, GitHub repo. The curriculum goes from imitation learning and RL all the way to Vision-Language-Action models and foundation models for robotics. Guest lectures from the co-founder of Physical Intelligence. The creator of Diffusion Policy. Pieter Abbeel. Dieter Fox. 12 weeks. Free. No signup. If you want to understand where robot intelligence is actually heading… this is the reading list the field is using right now. 📍[cvg.ethz.ch/lectures/Robot-L…] —— Weekly robotics and AI insights. Subscribe free: 22astronauts.com

Imagine you had the energy, metabolism, and mental agility of your younger self - without the aches and pains that come from aging. That’s what we’re attempting to build at @NewLimit and it will be one of the most important achievements in human history ever if it works. We're on the cusp of this being tractable given the rise of AI to test more hypotheses in-silico, along with the rapidly falling cost of single cell sequencing, etc. We announced our Series C and upcoming human clinical trials this week, so still a long, long way to go, with many risks. But with hard work we've got a shot at making this a reality.