Unconventional AI, a startup led by Naveen Rao, former head of AI at Databricks, released its first AI model on June 25, 2026, as an early demonstration of a new computing architecture the company says could reduce AI inference power consumption by up to 1,000 times.
The model, called Un-0, is an image-generation system built on an oscillator-based architecture that differs fundamentally from the chips used in conventional computing and large language models. The company published an accompanying research paper detailing how its team built a fully functional image-generation model using a software simulation of the new architecture, achieving performance comparable to existing diffusion models such as Stable Diffusion or OpenAI’s GPT Image 1.
“This is the ‘hello world’ of a new kind of computer,” Rao told TechCrunch. “Over the next year, you’re going to start seeing some pretty interesting news around this.”
The current version of Un-0 runs on a software simulation of Unconventional AI’s oscillator chips rather than physical hardware. The company plans to release chip schematics soon and intends to build a full inference stack from the ground up, eventually supplying compute capacity as a provider. “We will run AI models there, and we will have a network cable where prompts come in and inferences go out, but it’ll be done at 1/1000 of power,” Rao said.
Unconventional AI currently employs fewer than 50 people. The company’s focus on power efficiency stems from Rao’s view that energy availability will become a hard ceiling on AI development. “AI scaling is hard because of energy. It’s going to be the fundamental limit in the next few years. You just can’t go past it,” he said.
If the oscillator-based approach can be realized in physical hardware at scale, it could offer a path to meeting growing demand for AI inference without a proportional increase in power consumption — though much of the infrastructure required to reach that point is still under development.
Source: TechCrunch