China Won the AI Race Last Week
How would one build a plan to hurt Anthropic, OpenAI and Google’s model progress, and in the process handcuff the U.S. AI industry?
First, I would encourage Dario (Anthropic CEO), to do what he has done for the past 24 months - pursue his regulatory capture strategy. Dario regularly told the world how dangerous AI was every time someone put a microphone in front of him. Further, Dario regularly visited the Biden and Trump Admins and told them how dangerous AI is and that it must be regulated.
In doing so Dario made Government a player in the AI business. He handed them power. When Government got frightened by AMZN CEO Andy Jassy’s snitching about Anthropic’s Fable 5 model last week, it effectively pulled the model. That was not policy, that was a reaction.
Allow me to share with you what Government AI policy will be:
The Federal Government will roll out AI policy this year that will appoint AWS (AMZN), Azure (MSFT) and Google Cloud (GOOGL) as AI model gatekeepers (Maybe Musk’s cloud business (SPCX) will have a gatekeeper role as will Larry Ellison’s Oracle (ORCL)).
No AI model will be released commercially until the hyperscalers have blessed the models.
The hyperscalers are now in control (along with Government), and will eventually own the model companies, or what’s left of them in 2-3 years (the best and brightest will leave Anthropic and OpenAI).
This gatekeeper policy will ensure that the U.S. AI industry is a two or three-headed model beast shackled by the hyperscalers.
Upstart AI research labs will find it nearly impossible to compete given the regulatory barrier to entry.
Meanwhile, China-based AI labs are innovating at a crazy pace and making their models available via opensource. Those labs don’t have to play the regulatory capture game. There is no “AI gateway”. Those models are all in the public domain at a fraction of the cost of the U.S. models and it won’t be long (maybe a year or two), before they surpass Anthropic and OpenAI in capability.
My expectation is that opensource models will be banned this year or next, and that U.S. corporate customers will have to apply for exemption, which will cost time and money, which will be punitive to small companies like TEK2day. Banning opensource models will stifle innovation, will be unfair, and is anti-Capitalist.
The Republican/Democrat cabal that controls Washington D.C. is less Capitalistic than Beijing in many ways. Washington’s desire to stifle, control, and regulate the free market is evil. The regulation itself is harmful as it will stifle innovation and frankly the subject is not the Government’s business. Further, there is a money grab that will take place. Where do all of these regulatory fees go? To the Treasury’s balance sheet? Hah! You are a fool if you believe that. My guess is that only $0.10 of every regulatory $1.00 collected goes where it is supposed to go, the rest is washed through NGOs and off balance sheet. Regulation is a Government scam. This AI regulatory effort is going to bite the U.S. in the ass bigtime and China’s AI labs will 100% win the AI race. Anthropic, AWS and the Federal Government ensured that will be the case last week.