I saw OpenAI's news a few days ago about the establishment of an AI Deployment company today, and suddenly felt that this may be a track that has not yet entered mainstream view, or that the market has not fully priced it. It feels a bit interesting One current situation If it is the old iron that just entered American stock AI investment this month, it should be found that American stock is entering Hard mode from Easy mode. The best money to make in the past two years is essentially the PVE stage of AI Infra - GPU, HBM, optical module, nuclear power, data center, power, cooling, networking... The entire industry chain is almost fully recognized by the market. As long as it is related to AI data center, funds will be transmitted down the chain. But now the problem is also very obvious: this line has risen too completely. In many directions, there are even situations where profits have not yet been fully realized and valuations are being inflated first - for example, Cerebras, which opened two days ago and was aiming for a market value of 100 billion yuan, feels that Alpha is becoming increasingly difficult to find. The market has started to rebound in the past few days, which to some extent indicates one thing: the most easily profitable stage of the AI infras mainline is likely to have passed. II Something new? Now OpenAI is personally starting to do AI deployment. Do you want it? It is said that Anthropic is also doing similar things. I think this direction is very interesting because the market has not yet fully priced it, far from Despite the heated discussions about lobster Hermes and the like, I went back to China a few days ago to chat with some friends from non Internet giants in China. Most of their companies and enterprises used AI only as a bean bag, as a Demo and PPT. Even many corporate computers are prohibited from using ChatGPT and Claude Australia is not doing well either, as the vast majority of companies have not truly entered AI, and are far from achieving true workflow transformation. Now OpenAI/Anthropic is becoming more and more like Palantir+Accenture in the AI era. Directly dispatch deployment engineers to the site. Help enterprises change their organizational structure, workflow, software stack, and research and development processes. I used to often see comments saying that SaaS is dead, but in fact, most companies rely on inertia and are still stuck in the SaaS era. Someone/organization needs to push this AI transformation, and the market share of SaaS can really be replaced by AI agents+model calls AI Infra is now highly priced. But the real entry of AI into the internal of enterprises has just begun. The ceiling of this market is uncertain which one is larger than Infra, but it is definitely not low, after all, theoretically the upper limit of AI deployment is the entire global knowledge worker market. III Investment opportunities To be honest, I don't know because OpenAI and Anthropic are the ultimate players, which is currently a good thing for these two trillion dollar aircraft carriers? Another name that cannot be avoided is Palantir. Last year's superstar, one of the few AI application class rather than infrastructure class big white horses, has spent more than a decade verifying that the most difficult part of enterprise digitization is never the technology itself, but rather On site promotion, process improvement, overcoming organizational obstacles, breaking down data silos, understanding industry context, and truly running the system in a chaotic enterprise environment From this perspective, OpenAI and Anthropic are in a sense 'Palantirization'. So it's hard to say whether Palantir can reap this wave of dividends. Although it is an absolute leader in this direction, there is a key difference this time - Palantir represents the previous generation system integrator model. In this generation, model companies started to do it themselves. And Palantir's clients are all government, defense, and super large enterprises, so this round of OpenAI and Anthropic is targeting millions of ordinary enterprises This may also be a signal of industry stage transition, as AI is moving from simply selling shovels to transforming companies themselves. Perhaps the next Alpha investment opportunity will arise in this AI enterprise deployment track, giving birth to a form similar to Accenture in the AI era?