By Anne Tong, Kenrik Kai and Crystal Hu
Aalphabet’s San Francisco (Reuters), the biggest AI Scale client, plans to shorten scale ties after the news has broken up that the Meta opponent takes a 49% share in the launch of AI for data marking, said five sources familiar with the question.
Google had planned to pay a scale of AI around $ 200 million this year for learning data marked by a person who is crucial to the development of technology, including the complex AI models that Power Gemini, its Chatgpt competitor, said one source.
The search giant has already made talks with several AI rivals on AI this week as it seeks to displace much of this load, sources added.
The loss of a significant Scale business is coming as Meta takes a large share in the company, estimating it at $ 29 billion. The scale was worth $ 14 billion before the deal.
Scale Ai intends to maintain his business while CEO Alexander Wang, along with several employees, move to Meta. Because his main business is concentrated around several customers, he can suffer significantly if he loses key customers like Google.
In a statement, the AI Scale spokesman said his business, which extends to large companies and governments, remains strong as he is committed to protecting customer data. The company declined to comment on the specifics with Google.
The scale of $ 870 million in 2024, and Google spent about $ 150 million on AI’s AI services last year, sources said.
Other major technology companies that are Scale’s clients, including Microsoft, also retreat. Elon Musk’s XAI also strives to get out, one of the sources said. Openai decided to back off a few months ago, according to sources familiar with the question, although it spends far less money than Google. The OpenAi CEO said on Friday that the company would continue to work on a scale of AI as one of its many data providers.
Companies that compete with META in the development of avant -garde AI models are concerned that doing a scale business can expose its priorities to research and the rival’s roadmap, five sources said. By negotiating with scale AI, customers often share their own data as well as prototype products for which SCALE workers provide data labeling services. As Meta has now taken a 49% share, AI companies are concerned that one of their main rivals can acquire knowledge of their business strategy and technical plan.
Google, Microsoft and Openai declined to comment. XAI did not respond to a request for comment.
Rivals see openings
The bigger part of the AI revenue of the scale comes from the charging of manufacturers of generative models for providing access to a network of human trainers with specialized knowledge – from historians to scientists, some of whom are of doctoral degrees. People annotate the complex data sets used for “after train” AI models and as AI models have become more intelligent, the search for complex examples provided by man has increased and one explanation can cost as much as $ 100.