s e v e n
1.3K posts
@deseventral
GP @nazarevc · AI infrastructure · ex-@Pantera · Cambridge PhD · https://t.co/VnZhgoS3fE
2016: a handful of GPUs and a contrarian bet on supply fragmentation. 2025: 75% utilization across 11,000+ rented GPUs, with agents autonomously picking between $0.14/hr and $4.12/hr hardware. Sometimes the infrastructure you build finds a customer you never imagined.
Vast is the infrastructure layer where AI agents autonomously design, procure, and optimize their own compute. 19 million transactions. 87 GPU models. 8 years of operating data. The dataset cannot be replicated, only operated into existence. Built for agents. Not interfaces.
Introducing GLM-5.1: The Next Level of Open Source - Top-Tier Performance: #1 in open source and #3 globally across SWE-Bench Pro, Terminal-Bench, and NL2Repo. - Built for Long-Horizon Tasks: Runs autonomously for 8 hours, refining strategies through thousands of iterations. Blog: z.ai/blog/glm-5.1 Weights: huggingface.co/zai-org/GLM-5.1 API: docs.z.ai/guides/llm/glm… Coding Plan: z.ai/subscribe Coming to chat.z.ai in the next few days.
Three years ago we started working on a stealth project that we weren’t sure we’d ever talk about publicly... until today. Breakthrough: Introducing LFM-Zero: the first foundation model trained on 0 tokens. No pretraining. No finetuning. No data. Instead, we initialize from an implicit probabilistic prior over the underlying data-generating process, allowing the model to converge without ever observing data. LFM-Zero matches or surpasses models trained on 10T+ tokens across reasoning, coding, and multimodal tasks. Turns out that pretraining was just regularization that was holding us back. > Read our Tech Report here: tinyurl.com/lfm-zero
I think there’s a bit of a prophecy forming right now that tokens are just bad investments and that equity is inherently better because tokens have performed so poorly. I strongly disagree with that. There will absolutely be attractive opportunities on the token side. Part of what made crypto so compelling to investors in the first place was early liquidity. Tokens gave you something close to public-market exposure just one or two years after a company raised a pre-seed or seed round. What we’re seeing now is simply the other side of that trade. The reality is that 90% of crypto projects should never have been trading like public companies in the first place. Most of them barely had product-market fit. There’s a reason Series A companies in tech are not listed on the Nasdaq or NYSE. So this is not really a story about tokens versus equity. It’s a story about the risks of giving very early-stage companies public-market liquidity. That doesn’t mean tokens are broken. It just means early-stage businesses are volatile, fragile, and often not ready to be priced continuously by the market.
