I think China already disrupted the race enough to say they won.
Tech companies operate under the strategy of offering a subsidized service with the goal of capturing a monopoly share of the market, basically scale and worry about profiting later.
The plan was working as intended until the deepseek bomb dropped, and correct me if i am wrong since i am not an expert on the topic, they developed a way to train models with near same performance using data distilled from ChatGPT itself! This allowed deepseek to offer their API at several hundred times cheaper price on the already subsidized ChatGPT price, forcing openAI to drop their already unprofitable prices and capturing a share of the market instantly, there is also Qwen doing their thing.
The important thing imo is that the Deepseek method means that there will always be some degree of “cannibalizing” between these companies, when one invests a lot into training a model another one can copy it with a fraction of the investment.
I think China already disrupted the race enough to say they won.
Tech companies operate under the strategy of offering a subsidized service with the goal of capturing a monopoly share of the market, basically scale and worry about profiting later.
The plan was working as intended until the deepseek bomb dropped, and correct me if i am wrong since i am not an expert on the topic, they developed a way to train models with near same performance using data distilled from ChatGPT itself! This allowed deepseek to offer their API at several hundred times cheaper price on the already subsidized ChatGPT price, forcing openAI to drop their already unprofitable prices and capturing a share of the market instantly, there is also Qwen doing their thing.
The important thing imo is that the Deepseek method means that there will always be some degree of “cannibalizing” between these companies, when one invests a lot into training a model another one can copy it with a fraction of the investment.