By: Greg Reichow and Justin Selig

In the early eighties, the Apple II was considered the most powerful personal computer of its time. The majority of compute was dominated by large mainframes, massive computers that required an extraordinary amount of cooling due to the extreme amount of power utilized. Long before I became a Partner at Eclipse, I was an intern at Control Data, where the Chief Architect of the company’s supercomputers shared a glimpse of a skunkworks project that involved building an engineering workstation. It looked like a bigger version of a desktop computer, but contained a massive amount of processing power. I still remember the complex water cooling system required by not just the processor, but the entire main board of the computer. If this water system ever stopped, the computer would fry itself. In fact, as I received a demo of the computer, there was a failure of the cooling system and the inside of the computer completely melted down. 

My biggest takeaway from that seminal moment was how much energy compute required. Fast-forward to today: energy demands for compute have increased exponentially with large data centers on one end of the spectrum, all the way down to personal everyday devices like watches, phones, and security cameras. More and more of these edge devices get deployed every year and each iteration is even more power-hungry than the last. Given energy is the biggest design constraint for these small form factors, companies grapple with the need to cut features from applications to make them work on limited energy when conceiving new edge products. To solve this problem, many companies focus on bandaids to reclaim energy, like harvesting energy from alternative sources or searching for incrementally better battery chemistries. However, providing more energy is just one piece of the puzzle. The other piece involves figuring out how to use less total energy, while enabling more features on these devices. The second problem — decreasing the energy needs of compute — is complex and thus, many semiconductor companies have struggled to produce processors that reduce energy consumption since it is not obvious what levers to pull to do this with traditional processor architectures. As a result, the computing industry has been using about the same architecture since the 60s and is in dire need of a shake-up.

Today, the Eclipse team is excited to announce our partnership with Efficient Computer as it redefines energy-efficient computing with the introduction of a new computer processor that is up to 100x more energy-efficient than leading general-purpose CPUs on the market. With a $16M seed round led by Eclipse, Efficient Computer has created the novel Fabric processing architecture and software stack that delivers unprecedented levels of energy efficiency. Efficient’s processor marks the beginning of a new era of general-purpose edge computing that unlocks a wide range of applications previously restricted due to energy limitations. Accordingly, Efficient has the opportunity to revolutionize a $400B, growing, edge market. 

Efficient provides all the software, system hardware, silicon IP, and developer tools to deploy ultra low-energy edge processors. The company’s new architecture creates significantly more energy-efficient chips, delivering computing capabilities to form factors that were previously too small and energy-constrained. For example, Efficient’s processor could enable a smartwatch to last weeks between charges or preclude the need to send out field engineers frequently to check the status of enterprise IoT devices — all while supporting more sophisticated sensor data analytics because of Efficient’s general-purpose computing capabilities. 

Efficient’s processor architecture is differentiated in three ways: (1) orders-of-magnitude more energy efficiency, (2) orders-of-magnitude lower power, and (3) general-purpose programming capabilities. These value propositions map directly to the needs of current product developers focused on the far-edge and will address the demands of a wider class of developers as the proliferation of edge devices continues in years to come. 

The core IP for Efficient originally came from CTO Graham Gobieski’s PhD work at Carnegie Mellon. Eclipse Investor Justin Selig discovered Efficient through a former colleague who joined Efficient as a digital design engineer. We were impressed by CEO Brandon Lucia, a Professor of Computer Architecture at Carnegie Mellon and a pioneer in energy-constrained computing hardware and software. Brandon recruited world-class engineers, forming a team commendable for both its expertise and mutual respect. We noted, in addition to the caliber of the team, that their approach was both practical and scrappy. By the time we met Brandon and team for their seed round, they had already built and taped out a test System-on-Chip, alongside a fully-functioning and performant compiler stack. The progress that they made in a short time frame spoke volumes about the team’s ability to translate expertise into execution.

Powering a More Efficient Future

Highly inefficient general-purpose processors have slowed innovation, particularly at the edge. As the world continues to demand and develop applications for deployment on the edge, Efficient is the only platform that can deliver on the ultra low-energy, high-performance needs of these devices. More than just closing this gap, Efficient’s processor will empower organizations to reconsider how they build and deploy edge devices, consolidating the functionality of many disparate chips into one general-purpose chip. The Eclipse team is thrilled to partner with Efficient as it catalyzes innovation moving forward.

Congratulations to the entire Efficient team!

More on today’s news from Reuters

Follow Eclipse on LinkedIn for the latest on the Industrial Evolution.