GPU poor
In plain wordsSlang for researchers, startups or countries without access to the huge numbers of expensive chips needed to build the most advanced AI, as opposed to the GPU-rich big labs.
Lacking enough GPUsGPU, abbr.: A computer chip first designed to draw video-game graphics that turned out to be ideal for AI, because it can do thousands of simple calculations at the same time. or other acceleratorsAI accelerator, n.: Any chip built or tuned to speed up AI — graphics chips, Google's TPUs, phone 'neural engines' and many startup designs. Such chips do the heavy math of training and running models. to train or serve frontier-scale modelsfrontier model, n.: One of the most capable AI models in existence at a given time, built by a handful of leading labs at great cost, and the main target of AI safety rules.; also used as a noun ('the GPU-poor') for such researchers, startups, universities and countries, in contrast with GPU-rich labs and cloud providers that control tens or hundreds of thousands of top-end chips. The phrase treats concentration of training computetraining compute, n.: The total number of arithmetic operations performed to train a model. It is the standard yardstick for the size of an AI training effort, and some laws use it to decide which models face extra rules. as a defining fact of the field: the GPU-poor are pushed toward fine-tuningfine-tuning, n.: Taking an AI model that has already learned a lot and training it a little more on a smaller, focused set of examples so it gets better at one job or behaves a certain way. open models, efficiency research, small modelssmall language model, n.: A language model built to be compact enough to run cheaply and quickly, even on a laptop or phone, trading some knowledge and skill for speed, privacy and lower cost. and evaluation work. It is used both as self-deprecation and as a jab — the article that popularized it mocked GPU-poor groups for chasing leaderboardsleaderboard, n.: A public ranking table showing which AI systems score best on a test, updated as new results arrive — like a sports league table for models. with methods that would not matter at scale.
We're GPU poor, so we distill from open models and spend our budget on evaluation.
Origin
Popularized by Dylan Patel and Daniel Nishball in an August 2023 SemiAnalysis article, 'Google Gemini Eats The World – Gemini Smashes GPT-4 By 5X, The GPU-Poors', which sorted organizations into GPU-rich and GPU-poor; executives and researchers quickly adopted the label, often ironically. The underlying 'compute divide' between industry and academia had been studied since at least 2020.
Related
- GPUGPU, abbr.: A computer chip first designed to draw video-game graphics that turned out to be ideal for AI, because it can do thousands of simple calculations at the same time.
- training computetraining compute, n.: The total number of arithmetic operations performed to train a model. It is the standard yardstick for the size of an AI training effort, and some laws use it to decide which models face extra rules.
- moatmoat, n.: A lasting business advantage that keeps competitors out, like a castle's moat; in AI, the question of whether any company can stay ahead when rivals and free models catch up quickly.
- open-weight modelopen-weight model, n.: An AI model whose trained internal numbers are published for anyone to download and run on their own computers, even if the data and code used to make it stay private.
- H100H100, prop. n.: Nvidia's flagship AI chip of 2022–2024, so scarce and sought-after during the generative-AI boom that it became a unit of measure for AI computing power.
- compute governancecompute governance, n.: Using control over the computer chips and data centers that AI depends on — who can buy them, where they go, how they are used — as a way to oversee powerful AI.
Mentioned in
H100H100, prop. n.: Nvidia's flagship AI chip of 2022–2024, so scarce and sought-after during the generative-AI boom that it became a unit of measure for AI computing power., moatmoat, n.: A lasting business advantage that keeps competitors out, like a castle's moat; in AI, the question of whether any company can stay ahead when rivals and free models catch up quickly., wrapperwrapper, n.: A product built mostly on top of someone else's AI model, adding a simple interface or a few instructions; often used as a put-down implying the product is easy to copy., crackedcracked, adj.: Tech slang praising someone as exceptionally, almost unfairly, skilled; a 'cracked engineer' is a prodigy who ships remarkable work very fast., YOLO runYOLO run, n.: Slang for launching a big, expensive training run without the usual careful small-scale tests first — changing many things at once and trusting experience and instinct that it will work. YOLO means 'you only live once'.