Just wow, a city of Rio de Janerio post trained a 400B qwen model. We have Mumbai city corporation (BMC) at similar budget in USD. Not sure about Bengaluru budget after breakup into multiple corporations.

Compute: We were the first team in India to train at scale. We trained the sovereign models at the scale of ~3400 H100s. We are now putting serious capital behind the next step. India's first Blackwell cluster is now online and used by us, and we are building momentum towards operating 10s of megawatts in compute on Indian soil by 2027. Models: With Sarvam 105B, India's first sovereign model built from scratch, we showed that highly capable models can be trained here, independently. More importantly, the capability is now compounding across data, training, evaluation, systems, alignment, and deployment intuition. And we are scaling up to trillion-parameter class models, with larger runs built for coding, agents, and security. A coding model is coming soon... Inference: We already host our own models, with third party usage tripling in the last three months. We are soon taking live a production-grade token factory with the price, throughput, latency, reliability, and governance that banks, governments, enterprises, startups, and developers need for real systems. Products: Our products are now reaching India scale. Voice was our first wedge. It powers millions of interactions per day, doubling in the last three months, while we continue to optimise costs. Like voice, another modality at scale in India is documents, and we are hitting exponential growth of our new document intelligence product. Our fully managed agents product is live with enterprises and is being launched for all next month. Deployment: Most of the value in AI is unlocked in the last mile. We learned that by doing it across engagements in enterprises, government, and strategic sectors. Now we are turning that learning into a platform that allows every organisation to hill climb on its own use cases - whether it is building an agent, customising the harness, creating the data/tool backbone, or finetuning the model on custom data. Talent: Serious researchers are joining us across pretraining and RL, including people who have done meaningful work at the frontier. We are also starting our San Francisco office as the conduit for frontier AI ambition for India first, then the world.

A precedent is set. We have crossed the Rubicon and it’s hard to take either side at face value anymore.

Was a big advocate of asking founders to be in valley. When you don't have access to frontier models then it doesn't matter whether you are Bay or BLR if your AI team isn't 100% US citizens. It's a sovereign issue for countries like India, competitive survival for startups.

Export controls used to mean arms, then chips. Now it’s even a chatbot. And “foreign national” is half the talent that runs Silicon Valley. China has leading open-source models. India barely has a model. Now sit with that. This isn’t a desi VC problem. It’s a sovereign crisis.

Imagine if Anthropic was public.

Systems used to manage people now apply to the AI workforce: identity, role, access, budget, performance. Each AI coworker has a name, a mandate, a manager, and the tools to carry work from request to resolution. The existing workforce model, extended to AI. @atomicworkhq 👀 cc @khoslaventures

Today, we're launching the Atomicwork AI Workforce, and we believe this is one of the most significant shifts in how enterprises get work done. For the past decade, organisations have been building AI into their workflows. But the unit of operation that enterprises are actually built on has never been a task or a workflow. It's always been a person. Someone who owns a role, shows up with context, asks for help when they need it, and gets better over time. Every dimension your organisation already uses to manage people applies to the Atomicwork AI Workforce. Identity, job role, access, budget, performance. An AI coworker has a name, a defined scope, a human manager, and the tools to see work through from start to finish. The operating model is the same one your organisation has run on for decades. What's changed is that the technology to make this real has finally arrived, and for the first time, all the pieces are ready at the same time and they compose. We're not asking teams to rethink how they work. We're giving them a workforce that works the way they already do. Read more about what this means for your organisation Welcome to the #AtomicworkForce! @amnigos @kirandarisi @parsurv @AparnaChugh @rwmesh @leningali

To be fair, it’s more than Sequoia. VCs love minting unicorns. Last $5M goes in at a $1B cap, the press release is: “unicorn that raised $75M.” Founders will be fine. VCs are fine. The early employees who joined at that $1B mark when ARR is not even $2M? They’re the ones holding the bag. When you don’t agree to do this, VCs will label you as not ambitious enough 🤣

25 years ago was the first time I appeared on the Forbes Midas List. The technologies change. The game doesn’t. The biggest breakthroughs usually look crazy at first. The best founders build the future before consensus arrives.

Reminds me of Cloud Costs going out of control. Now it's Claude Costs 🤣 It's back again @dheeraj as you said :)

If anyone is wondering why Google Gemini AI mode still sucks. It's crazy that best company with great technology seems to under invest in the productisation of it :)

Delighted to see Gartner releasing ITSM platform migration guide for existing ServiceNow customers. Read more - lnkd.in/gKuArAiK Why this matters? Because the category is changing. @ServiceNow gave IT teams workflows. @atomicworkhq gives IT teams an AI workforce.

Story time: 1. What friendly VC's and close allies told me about building for IT teams. 2. Why I walked from more funding & acquisition offer$ 3. How ITSM will change for the next era beyond what incumbents like Workday and Salesforce are proposing. linkedin.com/pulse/itsm-hot-…

#74: #AIRadarDaily — @atomicworkhq For most of us, asking for help at work is a quietly frustrating experience. When your software access breaks or you need an HR document, you are forced to stop what you are doing, log into a clunky internal portal, and submit a ticket. Then, you wait in the dark. On the other side, brilliant IT and HR teams are drowning in a sea of repetitive queries, turning highly skilled professionals into stressed-out ticket routers. It is an exhausting cycle that drains the momentum from everyone involved. Atomicwork is stepping in to fundamentally rethink this entire experience. Founded by industry veterans @amnigos (Vijay Rayapati), Kiran Darisi, and Parsuram Vijayasankar, Atomicwork is an intelligent, AI-powered service management platform that completely automates enterprise support right where employees already work. The core of Atomicwork is a deeply sophisticated ""agentic"" architecture. They have built a coordinated team of specialized AI agents — an IT specialist, an HR operations agent, a security provisioner — all orchestrated by a universal conversational layer right inside Slack or Microsoft Teams. When an employee asks for software access, the AI instantly understands the context, checks company policy, runs the approval workflow, and provisions the software in seconds, all without generating a single manual ticket. What makes Atomicwork so special is its profound empathy for both the employee and the builder. It completely removes the bureaucratic friction of asking for help. For the IT and HR teams, it acts as an intelligent shield, deflecting the daily noise and giving them thousands of hours back to focus on building better systems instead of resetting passwords. It replaces the heavy, operational headache with quiet efficiency and speed. The market? Forward-thinking CIOs and global enterprises who realize that a great employee experience isn't a luxury — it is the absolute foundation of a fast, resilient business. Vijay, Kiran, Parsuram, and their team are true change agents. After spending years building world-class tools at places like Nutanix and Freshworks, they chose to tackle one of the most stubborn, deeply human pains in the corporate world: how we treat people when they need help. Watching these homegrown founders build a platform that restores time and dignity to the daily grind of work is exactly what it means to build for the larger, unfinished agenda. Let's celebrate the builders. w/ Jay Ingle & Dikshant Joshi #FutureOfWork #ProductNation

Overnight success, 10 years in the making. Conviction compounds. Perseverance is the moat. Congrats @amasad and team @Replit💜