How Infinity launched on X
Your Hardware. Frontier AI Performance.
What Infinity launched

Who builds the fastest AI? Whoever wins at automated research. That’s why top F100 companies like @Nvidia and @HP are coming to @infinity_ai_ . Infinity Autoresearch hit 2x token throughput for Nemotron 3 Super on the HP ZGX Fury (GB300). Here’s how we generate ∞ speed: https://t.co/n9UlrNWxYh
Transcript
How LaunchVideo.com rated the launch
LaunchVideo.com rating
About the launch
What is Infinity Autoresearch? Infinity builds Autoresearch, software that takes a hardware target and a serving workflow, then runs automated experiments over kernels, memory layout, and serving settings until inference throughput improves. The company describes itself as an automated AI research firm building the software layer that makes an AI chip ready for inference. Chip partners such as d-Matrix have used the approach to bring models up on new silicon. In a published case study, Infinity says it reached 92% of the theoretical peak on d-Matrix Corsair within 10 hours of first hardware access, and had three frontier models running end to end within 10 days. This launch applies that loop to hardware already in the field, together with HP. The focus is the HP ZGX Fury, a rack-mountable, office-deployable AI station based on the NVIDIA DGX Station platform and the GB300 Grace Blackwell Ultra Desktop Superchip. The box is built so several people can share it for local inference, or so a team can experiment and fine-tune in parallel. How did the Infinity launch perform on X? Jeremy Nixon, founder and CEO, posted the announcement on X on October 21, 2026, with the message "Your Hardware. Frontier AI Performance." The launch post reached 55K views, with 115 likes, 19 reposts, 25 comments, and 17 bookmarks, as recorded on the live LaunchVideo.com directory entry. The official collaboration announcement is dated October 20, 2026, the day before the post. This AI startup launch video is a technical results film, not a fundraising announcement. The seed round had already been disclosed in July. How did Autoresearch change performance on the HP ZGX Fury? Infinity and HP report that Autoresearch doubled concurrent workloads and doubled token throughput for NVIDIA Nemotron 3 Super, a 120.6B model, on the tested GB300 configuration, with no hardware change. The test used HP's existing vLLM benchmark, compared against an 82-concurrent-user baseline, and held a defined responsiveness threshold. Quality checks included NVIDIA SPEED-Bench, MMLU, and WikiText. The company says the system was given HP's baseline, ran 216 experiments, and reached the 2x result without speculative decoding, with the first gain in under 24 hours. Official materials also say throughput work continued after that first gain. The same launch thread reports 2.89x on Qwen 3.8 Max across eight B300s, 2.07x on Deepseek v4 Flash on one B300, and 2.19x on GLM 5.3 Flash across four B300s. Infinity says the optimizations span the inference stack, drawing on techniques used in vLLM, TensorRT-LLM, and SGLang, plus its own work. An earlier published H100 result, according to the company, showed its infy system delivering up to 34.3% more tokens per second than vLLM on Qwen3-8B under matched settings. Who is behind Infinity? Jeremy Nixon, a former Google Brain researcher, founded Infinity in San Francisco in August 2025. In July 2026 the company raised $15 million in seed funding at a $100 million post-money valuation, with significant participation from Touring Capital, along with Principal VC, executives at major chip companies, and researchers from OpenAI and Anthropic. Official materials also describe Ignition, Infinity's research agent for writing and optimizing low-level compute kernels on new silicon. Autoresearch, in this launch, is the same measurement and experiment loop applied to a workstation already sitting in an office, where twice the concurrency means more simultaneous users on one box. What can founders learn from this AI startup launch video? This AI startup launch video is a useful example of announcing a partnership with a measured result, not a slogan. The hook is simple, your hardware, frontier performance, and the body of the post is a test setup: a named model, a named machine, a baseline, a number of experiments, and a quality check. Founders studying the Infinity launch on X can also see the difference between a seed announcement and a proof announcement. The raise was already public. This post exists to show the system working on someone else's silicon. For a startup launch video example in infrastructure, the credible details are the baseline, the constraints you held constant, and the limits you still want the reader to know, such as results varying by model, workload, and configuration.
Tags
Questions
Frequently asked questions
The Infinity launch post reached 55K views on X, with 115 likes, 19 reposts, 25 comments, 17 bookmarks.
Infinity launched on X on October 21, 2026 with the message "Your Hardware. Frontier AI Performance.".
What is Infinity Autoresearch?
LaunchVideo.com rated the Infinity launch 3.00 out of 5: hook 4/5, editing / creativity 4/5, copy 3/5, sentiment of launch 3/5, distribution strategy 1/5.







