Ashutosh Singh

Trying to figure out how one guy can launch rockets, ruin Twitter, and still find time to sleep!

@durov Physics (with math)

📚 If you’re a student choosing what to focus on, pick MATH. It will teach you to relentlessly rely on your own brain, think logically, break down problems, and solve them step by step in the right order. That’s the core skill you’ll need to build companies and manage projects.

@ArthurConmy cool congrats Buddy Refer me 🎉

I'm joining Anthropic! I'll start work on aligning upcoming models as they’re trained Claude's capabilities are extraordinary. But like all models thus far, Claude isn’t aligned enough to safely delegate AGI development to I can't think of a better place to work on this at

THE MEMORY GIANTS ARE NOW WORTH A COMBINED $4.1 TRILLION 2016: Combined Market Cap $254B $204B - Samsung $26.2B - SK Hynix $24.1B - Micron $MU 2026: Combined Market Cap $4.1 Trillion $1.5T - Samsung $1.3T - SK Hynix $1.3T - Micron $MU

@MicronTech Ai doesn't Think as it's own it's the loading screen thinking... 💬

When an AI assistant “pauses” before responding, what is most often happening? #MicronTrivia

@durov Cool. Telegram on Wear OS is a nice touch.

Telegram is now on Wear OS. Owners of Galaxy Watch, Pixel Watch, Xiaomi Watch and other smartwatches — enjoy! ☕️

Just got access to Claude Fable 5. Anthropic's first Mythos-class model. Excited to test its reasoning, coding, research, and agentic capabilities on real-world workloads. @AnthropicAI The pace of AI progress is remarkable. #AI #Anthropic #ClaudeFable5 #LLM

Fable 5 is state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, scientific research, and vision. The longer and more complex the task, the larger Fable 5’s lead over our other models.

@NVIDIAKorea we want Gpus more Gpus 🙌

📣 한국의 통신망이 국가 AI 인프라로 진화하고 있습니다 SK텔레콤은 NVIDIA DSX 플랫폼을 기반으로 기가와트 규모의 AI 클라우드를 한국에 구축할 계획입니다. 🇰🇷 2027년 첫 번째 AI 팩토리가 가동을 시작해 한국 전역의 기업과 산업을 위한 소버린, 피지컬, 에이전트 AI 서비스를 지원하며, 아시아 전역으로의 확장을 목표로 하고 있습니다. 🔗 블로그 읽기: nvda.ws/3Ss5PtR

@wiz_io Blue is my favourite colour and cloud is Blue as well 💙

NEW: Introducing Wiz Cloud Cost 💸 We're bringing the power of Wiz to FinOps to help teams manage, optimize and govern cloud and AI spend - now Generally Available! Learn how Wiz Cloud Cost transforms your bottom line: wiz.io/blog/introduci…

Google and Anthropic are now paying Elon Musk's SpaceX a combined $2,170,000,000 per month for cloud compute capacity to run their AI services.

We raised $40M to make AI safe enough to trust. 🦢 Co-led by @WingVC and @MadronaVentures, with @Obviousvc, @Snowflake, @WeAreHRT, @SamsungNext, and @magaracvc. More fuel to deliver on the mission: empowering the world to use AI safely and securely. hubs.ly/Q04jdyhp0

@Tur24Tur @grok @grok is very smart

$300 → $99 for 3 months on SuperGrok Heavy. I claimed it. Not for the discount alone. I want @grok Build Beta in my daily workflow. 16x agents, heavy limits, early access. Let’s see if it keeps up with real offensive work.

@GoogleVRP I genuinely believe AI is exceptionally good at identifying vulnerabilities and scaling security analysis across massive codebases far faster than any human team alone.

📣Blast from the past📣 This post takes us back to a flaw discovered in 2010: while technology has advanced, the general story of how the flaw was detected is still a great example of effectively identifying and remediating a security issue. bughunters.google.com/blog/bit-rot-d…

Watch me control my computer with just my voice. This is the future of operating systems. No hands. GPT-Realtime 2.0 is very, very underrated. Demo:

SpaceX has almost finished writing V1.0 of an in-house AI training stack in C that exact-maps to 220k GB300s with 800G NICs, making heavy use of pipeline parallelism and getting as close to bare metal as possible. The potential speed improvement vs JAX for large training runs is over an order of magnitude.

Jane Street AI Engineer revealed how they trained their own LLM for trading to make $22.5B/year 16 minutes. free. straight from tier-1 quants. bookmark & watch - this is the most honest "AI inside a hedge fund" talk ever published. forget the "AI trading bot" YouTube grifters. This is the real inside view: data, training, evals, integration. then start building your own bot using post below.

Why did xAI hand over a 220,000-GPU cluster to Anthropic? The technical backdrop to xAI's decision to hand Colossus 1 over to Anthropic in its entirety is more interesting than it appears. xAI deployed more than 220,000 NVIDIA GPUs at its Colossus 1 data center in Memphis. Of these, roughly 150,000 are estimated to be H100s, 50,000 H200s, and 20,000 GB200s. In other words, three different generations of silicon are mixed together inside a single cluster — a "heterogeneous architecture." For distributed training, however, this configuration is close to a disaster, according to engineers familiar with the setup. In distributed training, 100,000 GPUs must finish a single step simultaneously before the cluster can advance to the next one. Even if the GB200s finish their computation first, the remaining 99,999 chips have to wait for the slower H100s — or for any GPU that has hit a stack-related snag — to catch up. This is known as the straggler effect. The 11% GPU utilization rate (MFU: the share of theoretical FLOPs actually realized) at xAI recently reported by The Information can be read as the numerical fallout of this problem. It stands in stark contrast to the 40%-plus MFU figures achieved by Meta and Google. The problem runs deeper still. As discussed earlier, NVIDIA's NCCL has traditionally been optimized for a ring topology. It works beautifully at the 1,000–10,000 GPU scale, but once you push into the 100,000-unit range, the latency of data traversing the ring once around becomes punishingly long. GPUs need to churn through computations rapidly to keep MFU high, but while they sit waiting endlessly for data to arrive over the network fabric, more than half of the silicon falls into idle. Google sidestepped this bottleneck with its own custom topology (Google's OCS: Apollo/Palomar), but xAI, by my read, has not yet reached that stage. Layer Blackwell's (GB200) "power smoothing" issue on top, and the picture comes into focus. According to Zeeshan Patel, formerly in charge of multimodal pre-training at xAI, Blackwell GPUs draw power so aggressively that the chip itself includes a hardware feature for smoothing power delivery. xAI's existing software stack, however, was optimized for Hopper and does not understand the characteristics of the new hardware; when it imposes irregular loads on the chip, the silicon physically destructs — literally melts. That means the modeling stack must be rewritten from scratch, which in turn means scaling is far harder than most of us imagine. Pulling all of this together points to a single conclusion. xAI judged that training frontier models on Colossus 1 simply was not efficient enough to be worthwhile. It therefore moved its own training workloads wholesale onto Colossus 2, built as a 100% Blackwell homogeneous cluster. Colossus 1, on the other hand — whose mixed architecture is far less crippling for inference, which parallelizes more forgivingly — was leased in its entirety to an Anthropic that desperately needed inference capacity. Many observers point to what looks like a contradiction: Elon Musk poured enormous capital into building Colossus, only to hand the core asset over to a direct competitor in Anthropic. Others read it as xAI capitulating because it is a "middling frontier lab." But these are surface-level reads. Look at the numbers and a different picture emerges. xAI today holds roughly 550,000+ GPUs in total (on an H100-equivalent performance basis), and Colossus 1 (220,000 units) accounts for only about 40% of the total available capacity. Colossus 2 — built entirely on Blackwell — is already operational and continuing to expand. Elon kept the all-Blackwell homogeneous cluster (Colossus 2) for himself and leased out the older, mixed-generation Colossus 1. In other words, he handed the pain of rewriting the stack — the MFU-11% debacle — to Anthropic, while keeping his own focus on training the next generation of models. The real point, then, is this. Elon's objective appears to be positioning ahead of the SpaceXAI IPO at a $1.75 trillion valuation, currently floated for as early as June. The narrative SpaceXAI now needs is that xAI — long the "sore finger" — is not merely a research lab burning cash, but a business with a "neo-cloud" model in the mold of AWS, capable of leasing surplus assets at high yields. From a cost-of-capital perspective, an "AGI cash incinerator" is far less attractive to investors than a "data-center landlord generating cash." As noted above, the most important detail of the Colossus 1 lease is that it is for inference, not training. Unlike training, inference requires far less tightly synchronized inter-GPU communication. Even when the chips are heterogeneous, the workload parcels out cleanly across them in parallel. The straggler effect — the chief weakness of a mixed cluster — is essentially neutralized for inference workloads. Furthermore, with Anthropic occupying all 220,000 GPUs as a single tenant, the network-switch jitter (unanticipated latency) that arises under multi-tenancy disappears. The two sides' technical weaknesses end up complementing each other almost exactly. One insight follows. As a training cluster mixing H100/H200/GB200, Colossus 1 was an asset that could only deliver an MFU of 11%. The moment it was handed over to a single inference customer, however, that asset transformed into a cash-flow asset rented out at roughly $2.60 per GPU-hour (a weighted average of the lease rates across GPU types). For xAI, what was a "cluster from hell" for training has become a "golden goose" minting $5–6 billion in annual revenue when redeployed for inference. Elon's genius, I would argue, lies not in the model but in this asset-rotation structure. The weight of that $6 billion becomes clearer when set against xAI's income statement. Annualizing xAI's 1Q26 net loss yields roughly $6 billion in losses per year. The $5–6 billion in annual revenue generated by leasing Colossus 1 to Anthropic, in other words, almost perfectly hedges xAI's loss figure. This single deal effectively pulls xAI to break-even. Heading into the SpaceXAI IPO, this functions as a core line of financial defense. From a cost-of-capital standpoint, if the image shifts from "research lab burning cash" to "infrastructure tollgate stably printing $6 billion a year," the entire tone of the offering can change. (May 8, 2026, Mirae Asset Securities)

Happy World Quantum Day! Today’s #GoogleDoodle represents the potential of the qubit. We’re harnessing its unique quantum properties to build systems capable of solving problems that remain intractable for classical computers → goo.gle/4ckighH