Low-energy chip startup Efficient Computer closes $97 million in financing

Low-power chip startup Efficient Computer Co. announced today that it has closed a $97 million funding round, the second major investment it has made this year.

The round was led by TQ Ventures and included Eclipse, Union Square Ventures, Giant Ventures, Triatomic Capital, TO Capital, TF Capital, Mana Ventures, Toyota Ventures, Overmatch and Borderless. This brings the company’s total amount raised to date to $650 million.

Efficient, which closed a $60 million round in February, is in the process of bringing to market an entirely new, energy-efficient computer chip based on a dataflow architecture. It is claimed to reduce power consumption compared to the x86 chip architecture from companies like Intel Corp. and Advanced Micro Devices Inc. can potentially reduce costs by a factor of 100.

Efficient says its chips minimize energy consumption to the point where they could theoretically power devices for months or even years. If Efficient can truly do this, its chips will prove extremely useful in today’s power-hungry artificial intelligence environments.

Modern central processing units and graphics processors are extremely inefficient in terms of energy consumption, prioritizing high performance, low latency, and massive throughput over energy conservation. As a result, they are constrained by architectural overheads, including complex control logic, high-speed data movements, and the need to maintain precise execution status.

In a blog post announcing today’s round, Brandon Lucia, co-founder and CEO of Efficient, said the specialty processors used today to accelerate AI workloads are a “devil’s bargain” because they sacrifice programmability and adaptability to increase their efficiency and speed. They are only useful for a very limited subset of tasks, namely AI inference, meaning they are useful for general calculations.

Efficient’s data flow architecture eliminates all unnecessary data movement and architectural overheads inherent in today’s CPU and GPU architectures. By intelligently distributing workloads and linking instructions to reflect the application’s data flow, dramatic increases in performance per watt can be achieved without sacrificing performance or sacrificing programmability.

Dataflow chips have been designed for decades in the scientific literature but have never gained commercial traction because of the difficulty of programming them for the wide variety of tasks that chips from Intel, AMD and Nvidia can handle. However, the desire for more energy-efficient AI processors has sparked renewed interest in the idea.

Lucia says Efficient took the basic concepts of dataflow architecture and then went back to the drawing board to design both the chips and the software tools needed to make them as adaptable as today’s general-purpose processors. Although the company ultimately wants to make larger chips that can power data center servers, its initial target market is more modest.

Its first chips are intended to power small robots and autonomous drones that run on batteries instead of a power supply. “We are threading the needle where we can program easily, quickly and efficiently,” Lucia said in an interview with Reuters. “When you build an AI system, the system ends up doing much more than just two small nano-optimized AI algorithms.”

Efficient hasn’t revealed the names of its customers or details of its revenue, but Lucia insisted the company is seeing “overwhelming demand” for its first product, the Electron E1 chip. Funds from today’s round will help Efficient increase production volumes and increase deliveries to its customers.

Andrew Marks, partner at TQ Ventures, said he supports Efficient because the need for greater energy efficiency goes beyond just running AI models. “What sold us was Efficient’s ability to develop both the hardware and the software and turn this breakthrough into a business,” he explained. “Not only have they ramped up production four times over, but they are already shipping chips to customers in large quantities.”

Image: Efficient computer