Sources: OpenAI, Google, and Anthropic are all seeing diminishing returns from costly efforts to build new AI models; a new Gemini model misses internal targets
Three of the leading artificial intelligence companies are seeing diminishing returns from their costly efforts to develop newer models.
Bloomberg
Context & Ripple Effects
Earlier coverage had already shown how capital-intensive frontier AI had become, with well-funded AI startups struggling against Big Tech rivals. That backdrop makes reports of weaker returns at the leading labs more consequential than an isolated product setback.
OpenAI, Google, and Anthropic face greater pressure to justify the cost of successive model-training efforts when reported capability gains are becoming harder to obtain.
Google must address the gap between a new Gemini model's internal performance targets and its results, raising the execution stakes for its model roadmap.
Second-order effects
If frontier-model gains cost more to achieve, product teams and enterprise buyers gain leverage to prioritize efficiency and practical performance over headline model advances.
If diminishing returns persist, frontier AI competition may shift from a race centered on ever-larger training runs toward an industrialized model focused on efficient deployment, integration, and cost control.
The advantage of the best-funded labs could become less about funding alone and more about whether they can convert expensive model development into products with durable distribution and economics.
The trend: Frontier AI is moving from a scale-first race toward a price-performance and commercialization contest as the marginal payoff from new training runs comes under scrutiny.
Lots of good information here. “After years of pushing out increasingly sophisticated AI products at a breakneck pace, three of the leading AI companies are now seeing diminishing returns from their costly efforts to build newer models.”
Bloomberg reports that OpenAI, Google, and Anthropic are having trouble making better models, seeing less progress despite higher costs - Two people familiar with OpenAI's Orion project say the September 2024 model fell short when trying to answer coding questions it hadn't been …
This is big: we learned 3 of the hottest AI companies — OpenAI, Google and Anthropic — are all struggling to build their next frontier AI models. Read @shiringhaffary, @dinabass, @byJuliaLove and I on what's happening and what's next. Here's a gift link! https://www.bloomberg.com…
Idk if these rumors are true but, if they are, even more reaosn that 2025 will need to be the year that AI *interfaces* help LLMs cross the chasm. Existing frontier models are already so powerful & underutilized. @cursor_ai is the only (truly) AI-native product I've seen. [image]
“After training it, Anthropic found 3.5 Opus performed better on evaluations than the older version but not by as much as it should, given the size of the model and how costly it was to build and run, one of the people said.” [image]