100% Agree! A few questions I keep chewing on, which you've probably gotten before:
If inference commoditizes, why doesn't training? RL recipes, distillation pipelines, environment tooling — these feel like they could go public within months, the same way kernel tricks did. What makes the training layer hold?
On frontier gravity: a lot of GPT-3.5 finetuning investment got wiped out overnight by GPT-4 zero-shot. The thesis seems to assume domain-specific evals stay ahead of base model progress. Is that territory actually big enough to support the 90% claim?
If a big motivation for training your own model is protecting data/IP, doesn't using an external training provider cut against that?
Curious how you draw the trust boundary so customer data and evals compound for the customer, not for the provider.
@ypatil125 Makes sense boss. The inference vendors are all adding post training, inevitable that it also goes the other way
At Applied Compute, we believe Training wins Inference.
In the limit, inference alone is a commodity: providers serve the same models, compete on price, and race toward thinner margins.
Training changes that.
Once an AI workload reaches scale, serving a generic model stops making sense. Companies will fine-tune, distill, and continually optimize models around their own data, product, economics, and definition of quality.
That is the real advantage of open-weight models: you have the weights.
You can make a model better on your evals, faster for your workload, and cheaper per token. Over time, companies will choose the best model for their specific task - not the best generic model.
We believe that, in the long run, more than 90% of inference on open-weight models will come from trained variants rather than untouched base models.
Training also de-commoditizes inference!
A provider serving the same off-the-shelf model as everyone else has little room to differentiate. But a company will pay more for a trained version of GLM-5.2 that materially outperforms the base model on the evals that they care about.
The providers that help companies train differentiated models (Applied Compute) will be best positioned to serve them in production, too. Training and inference will converge into one continuous system.
The long-term end state is continual learning: models that improve from proprietary data, real-world usage, and direct feedback.
The companies that close this loop fastest will build products that get better with every interaction and compound their lead over time.
