CEOs of many star AI companies are jumping ship intensively, and the inflection point of artificial intelligence is coming?

The CEO of Adept jumped to Amazon, the CEO of Inflection left the company to join Microsoft, and the CEO and president of Character.AI ran away to Google. In just 5 months, executives from 3 star AI startups have switched to large companies. The tech giants, on the other hand, continue to invest heavily as if the end of victory is in sight and will not give up. The momentum of artificial intelligence soaring seems to have reached a crossroads.

Another CEO of a star AI company has jumped ship

On August 2, Character.AI announced that CEO Noam Shazeer and President Daniel De Freitas, along with several members of the research team, have left the company to return to their old club Google.

In 2021, Google employees Noam Shazeer and Daniel De Freitas left Google because they were dissatisfied with Google's bureaucracy. The duo then founded Character.AI, with Noam as CEO and Daniel as president. Character.AI is an AI-based chatbot platform that allows users to interact and have conversations with different AI characters. These AI personas are designed to provide personalized help and support at every moment of the user's day.

Since its establishment, the company has grown rapidly, with more than 4 million mobile users at its peak. In 2023, the company completed a Series A financing with a financing amount of $193 million and a valuation of $1 billion. But as user growth slowed and the continued high cost of model training continued, the company finally decided to "sell" Google.

According to the announcement released by the Character.AI, the company's goal is to achieve personalized superintelligence, which requires a full-stack approach. Companies must pre-train the model and train it later to support the unique experience of the Character.AI while building a product platform that can reach users around the world. But the landscape has changed over the past two years, and more pre-trained models are now available. In light of these changes, the Company sees greater use of third-party LLMs in conjunction with its own models as an advantage. This allows the company to devote more resources to follow-up training and create new product experiences for its growing user base.

Noam is excited to leave the company he founded and return to Google, and Character.AI will get better and better in the future. He is proud of Character.AI achievements over the past three years. He believes that funding from Google's non-exclusive licensing agreement, coupled with a fantastic Character.AI team, will enable Character.AI to continue to be successful in the future. Google said Noam would join the DeepMind research team, but did not specify his or De Freitas' exact role.

In addition, the company said that Google agreed to pay Character.AI for licensing its models and to hire CEOs and researchers at a high price. However, the company did not disclose the exact amount. Character.AI also informed employees that previous investors would receive a buyback of $88 per share. That's about 2.5 times the value per share in Character.AI's Series A funding round in 2023, when the company was valued at $1 billion.

It is reported that Character.AI will switch to using open-source models, such as Meta's Llama 3.1, to support its products, rather than its internal models.

Will the AI bubble burst?

Character.AI founders are not alone in leaving and moving on to larger companies.

In June, Adept co-founder and CEO David Luan announced that he would be joining Amazon, along with several of Adept's co-founders and other employees. Adept is a startup that specializes in developing AI-based "agents" to accomplish a variety of software-based tasks.

In March, Mustafa Suleyman and Karén Simonyan, co-founders of AI startup Inflection AI, announced that they were joining Microsoft. Suleyman is responsible for Microsoft AI, Microsoft's newly formed consumer AI division, while Simonyan will join the company as chief scientist. Just a year ago, Microsoft led a $1.3 billion investment in Inflection AI, valuing it at about $4 billion (nearly 30 billion yuan), making it one of the largest funding rounds among AI startups.

In addition, Stability AI, which created the Stable Diffusion series of models, is also facing an unprecedented financial crisis. The once-$1 billion valuation, but 180 employees, is considering a sale and is actively approaching potential buyers.

Commenting on the setbacks suffered by many of the industry's star startups, Marcus, a professor at New York University and a best-selling author, believes that it's time to burst the bubble of generative AI.

But at the same time, big tech companies are still confident. Sundar Pichai, Google's chief executive, said the company would choose to overinvest rather than miss out on potential AI revenue opportunities: "For us, the risk ·of underinvesting is far greater than the risk of overinvesting, even if it turns out that we are overinvesting." ”

Microsoft CEO Satya Nadella may be the most optimistic of · all. He says he's seeing strong demand signals from customers: "Microsoft 365 Copilot is our best Office 365 or M365 suite, and a lot of customers come back after using it. ”

And the investment institutions are equally optimistic.

According to The Information's latest statistics, the generative artificial intelligence (GenAI) craze is far from subsiding. GenAI startups raised a record $12.2 billion in the second quarter, surpassing the record set in the first quarter of 2023. The previous record was largely due to Microsoft's $10 billion investment in OpenAI, and this time it was Musk's xAI that raised $6 billion in May.

But even excluding xAI, GenAI startups have seen a more than 85% increase in funding compared to the same period last year. At the same time, the number of companies that have secured funding has exceeded that of any quarter since records began in 2021 — with as many as 55.

Some people resigned and returned to their hometowns, and some people rushed to the field in the starry night. Who's right and who's wrong, the answer shouldn't be too far away.

Editor-in-charge: Yue Yanan

Proofreading: Ran Yanqing