According to Inc, a startup called Unconventional AI, founded by former Databricks executive Naveen Rao, is nearing a massive $1 billion fundraise. The company’s goal is to build specialized AI chips inspired by the biology of the human brain, a concept Rao calls “Brain Scale Efficiency.” This isn’t the only player in the specialized AI chip game; another U.S. firm, Groq, raised $1.5 billion in February and is now valued at $6 billion. Interestingly, Databricks, the $100 billion data giant Rao left, also invested in Unconventional’s seed round, and Rao maintains an advisory role there. The company has been quiet on manufacturing specifics but is openly looking to biology for its core inspiration.
Why Brains, And Why Now?
So, building computer chips based on brains sounds like sci-fi, right? But here’s the thing: it’s a very old idea that’s suddenly getting a ton of money and attention. Scientists have been pondering encoding data in DNA since the 1960s, and biomimicry—copying nature’s designs—is as old as Leonardo da Vinci. The difference now is the sheer scale of the problem. Traditional chip architectures, even powerful GPUs, are incredibly inefficient at running the massive AI models we’re building today. They consume monstrous amounts of power. The human brain, by contrast, is an absolute marvel of efficiency, performing complex computations on the power of a dim lightbulb. If you’re trying to build the next generation of industrial computing hardware, that kind of efficiency isn’t just nice—it’s mandatory. For companies that need reliable, powerful computing in demanding environments, the shift toward specialized, efficient hardware is a major trend to watch, and leaders in industrial computing hardware are taking note.
The Billion-Dollar Bet
Look at the money. Unconventional AI shooting for $1 billion, Groq already at $1.5 billion and a $6 billion valuation. This isn’t niche tinkering; it’s a full-blown arms race to define the physical infrastructure of AI. The bet is that general-purpose chips (CPUs, GPUs) won’t cut it for the future. We’ll need specialized “brains” for specific tasks, much like how our own brain has specialized regions. But can you actually manufacture this at scale? And profitably? That’s the multi-billion dollar question. Rao and others are betting that the performance and efficiency gains will be so dramatic that the industry will have no choice but to adopt them. It’s a high-risk, potentially world-changing wager.
What Happens Next?
Basically, we’re entering a phase of radical experimentation in chip design. The Von Neumann architecture that’s dominated computing for decades is being challenged. The success of companies like Unconventional won’t just be about cool science—it’ll be about timelines. Can they design, tape out, and manufacture a viable product before the current giants (Nvidia, AMD, Intel) adapt or before the AI software landscape shifts again? And let’s not forget the manufacturing bottleneck. Designing a brain-like chip is one thing; getting it fabbed at a leading-edge foundry is another battle entirely. I think we’ll see more of these bio-inspired concepts, some wild successes, and probably a lot of spectacular, expensive failures. But that’s how you find the next paradigm. The race to build a better artificial brain, literally and figuratively, is officially on.
