A startup called Flourish has raised $500 million and reached a reported valuation of $2.5 billion in 2026, with the goal of developing a new approach to artificial intelligence based on how biological neurons actually work. Jeff Bezos is among the key backers, along with Lux Capital and Google Ventures.
The company was founded by Thomas Reardon IV — known simply as Reardon — a former Microsoft programmer who later earned a doctorate in neuroscience from Columbia University. His career includes helping build Microsoft’s first web browser, founding and selling a wireless technology company, and developing a mind-control wristband that was acquired by Meta, where he worked for six years.
Reardon’s co-founder is a former Amazon executive named Williams, whom he knew from their time at Microsoft. Greg Wayne, a longtime researcher at DeepMind who leads Project Astra — Google’s AI assistant initiative — also joined as an early recruit. DeepMind CEO Demis Hassabis negotiated an arrangement allowing Wayne to split his time, dedicating 20 percent to Flourish while keeping his DeepMind role.
Bezos initially contributed $50 million after reading a two-page pitch from Williams, then nearly doubled his stake — later saying he would have given more if asked.
The motivation behind Flourish stems from Reardon’s dissatisfaction with how major companies, including Meta, have approached AI development. He and his team argue that current large language models, despite being called neural networks, do not actually replicate brain-like processes. UC Berkeley computer scientist Ben Recht, a Flourish adviser, noted that neuromorphic approaches to software predate the LLM era but were largely set aside as LLMs became dominant.
Both IBM and Intel have previously released neuromorphic chips inspired by brain architecture, but Flourish is pursuing a broader reinvention of AI itself. Whether the approach can succeed remains an open question — Wayne acknowledged he was uncertain the team could achieve its goal, but believed the work “would lead to interestingness.”
Source: WIRED