Generative AI Is an Engineering Disaster -- A shockingly inefficient trillion-dollar project

The problem with generative AI, in the industry’s own jargon, is that it does not scale. The cost of growing from, say, a thousand users to a million is a key factor that venture capitalists examine when they evaluate start-ups. They want to see that the cost of adding each new user decreases over time, so that the company can support millions of users and make increasing profits. This is achieved partly through the careful engineering of computer systems that can efficiently handle more users who want to post photos, hail Ubers, or stream music.

With generative AI, the work of building efficient, scalable systems has not been done. And the problem is exacerbated by the ever-larger generative-AI models, which have grown from 175 billion parameters in 2020 to more than 1 trillion today, according to independent estimates (the actual sizes of the models powering products such as Claude and ChatGPT are secret). The large in large language model should not be a selling point. But the industry’s observation that bigger models tend to outperform smaller ones has given rise to a totemic belief in “scaling laws” that suggest any problem can be solved by simply making models bigger. “Maybe with 10 gigawatts of compute, AI can figure out how to cure cancer,” OpenAI CEO Sam Altman wrote on his blog in September.

Yet the returns are diminishing. The bigger an AI model is, the less it improves with each added parameter, and so it must be made bigger at a faster rate just to sustain steady progress. I asked a few AI researchers whether they could name any other real-world software that scales so poorly. None of them could think of any. Even outside the world of software, it’s hard to find a comparable example, given that economy of scale is the principle that has made light bulbs, cars, and clothing so affordable. By economic and engineering measures, generative AI might be the worst technology ever deployed

$wall removed: https://removepaywalls.com/https://www.theatlantic.com/technology/2026/07/generative-ai-engineering-disaster/687901/

if by shockingly you mean blantantly predictable and completely obvious

Might be a hot take here but personally I view LLMs as nice piece of technology that is being used in very "inefficient" and "wrong" ways.

We have technology that can itrepet meaning and extract facts from natural language at striking accuracy which has been problem for symbolic processes for decades.

But we are using the technology to create hallucinating agents, that follow rules probabilistically and are vulnerable to social engineering through pretty much all possible channels.