This is a modus operandi that we fear to see: a large company concluding an agreement with a smaller one to poach part of its staff. Recently, the start-up Character.ai paid the price. The latter has developed a platform where it is possible to create or simply chat with robots supposed to embody living or dead, real or fictitious personalities. Only here: its founders have just been poached by Google as part of a transaction worth 2.7 billion dollars, as indicated by the Financial Times.
Based in San Francisco, the start-up was born in the midst of the generative AI boom. Behind her, two veterans of the Mountain View firm: Daniel De Freitas and Noam Shazeer. The two previously left the tech giant after it refused to release its AI-powered chatbot.
Noam Shazeer is also one of eight Google scientists behind the paper on the “Transformer” architecture for language processing, which kicked off the generative AI revolution. As of March 2023, the company even raised $150 million from prestigious Silicon Valley venture capital funds, such as Andreessen Horowitz. A round of financing which allowed it to exceed the symbolic mark of a billion dollars in valuation, only a year and a half after its launch.
A previous agreement took place this summer
As part of the agreement with Google, the firm has therefore rehired Noam Shazeer and Daniel De Freitas. In August, the firm already hired 20% of Character.ai’s staff to join its DeepMind AI branch. This represents around thirty employees. In addition, the giant paid $2.7 billion for a single license to the start-up’s models to date, without access to future technologies, according to people familiar with the transaction.
Dominic Perella, the company’s new interim CEO, hopes the Google deal won’t raise antitrust concerns because it plans to operate in the same market. “We continue to do research on AI,” he declared. “We still own all of our technology, we have almost all of our staff and we continue to grow.”
A risk of being caught by antitrust regulators
In this type of case, the risk of arousing the curiosity of regulators is never far away. In the past, many agreements have proven that certain alliances are difficult to create, like Microsoft’s with OpenAI, worth $13 billion. More recently, another operation has the authorities twitching: Microsoft’s $650 million deal in March to hire Inflection’s director, Mustafa Suleyman, and other members of the start-up’s staff.
The British competition regulator therefore carried out an investigation, calling it a “merger situation” before eventually abandoning its research. The European regulator, also attentive to this operation, ended up also stopping following a court decision from the Court of Justice of the EU. The $4 billion deal between Amazon and the start-up Anthropic suffered the same fate.
What about the future of Character.ai?
With the 2.7 billion dollars resulting from the transaction with Google, Character.ai decided to buy back the shares of its investors and distributed them among the employees within the framework of a cooperative, a “very unique and perhaps unprecedented structure in Silicon Valley”, Dominic Perella told the FT. The latter has a stake of less than 10%. The rest of the funds must be used to run the business for 18 months. It is not impossible that it will seek to raise funds again and enter into other licensing agreements.
Its founders – from Google – were the guarantor of the company. Without them, the future of Character.ai could be in jeopardy. Around a hundred employees remain to date: they will focus on improving existing products – chatbots – rather than on the development of AI models. Dominic Perella told the Financial Times that the start-up had largely dropped out of the LLM race to better-funded rivals such as Microsoft-backed OpenAI, Amazon and Google.
Focus on an already finished product rather than developing new models
“It has become incredibly expensive to train frontier models, even with a very large start-up budget”he indicated. The start-up now boasts a monthly active user base of 20 million, which has doubled year-on-year, with a predominantly younger user base, aged 13 to 25.
Its revenues come mainly from subscriptions. At the same time, Character.ai has just announced the recruitment of Erin Teague as product manager. The latter previously worked at Google as a product manager on the Gemini models and on YouTube.
Tech giants are crushing the competition
Big Tech regularly attacks smaller companies, that’s a fact. The risk, at a time when everyone is trying to develop their own generative AI model, is to see these start-ups and their solutions caught in the cracks launched by Amazon, Apple, Google, Meta or even Microsoft. Crushing the competition would offer them a avenue to develop their own models and their own vision of generative AI. Character.ai had also already been the subject of takeover interest, notably from Meta, without any agreement ultimately being concluded.
Selected for you
Related News :