The world of computer chips presents a remarkably wide array of options, potentially exceeding the variety found in a typical grocery store’s snack selection. The chip industry is currently experiencing significant disruption due to the diversity of computing environments – including data centers, the cloud, edge computing, mobile devices, and the Internet of Things – alongside differing cost considerations and performance demands. This is leading to a reshuffling of leadership and the emergence of new contenders in specialized areas.

Numerous new chip companies, such as Cerebras, SiFive, and Nuvia, have received venture capital funding in recent years. However, Flex Logix established itself somewhat earlier in the field. Founded in 2014 by Geoff Tate, a former founder of Rambus, and Cheng Wang, the company has secured a total of $27 million in investment from Lux Capital, Eclipse Ventures, and Tate himself.

Flex Logix aims to enable AI processing capabilities at the edge of computing networks. This involves providing technologies that integrate artificial intelligence into devices like medical imaging systems and robotic equipment. For edge applications, both processing capability, as well as physical size and cost, are crucial factors. More efficient chips are simpler to integrate into products, and price limitations can influence the expense of individual components.

Initially, the company concentrated on the development and licensing of intellectual property related to FPGAs – Field Programmable Gate Arrays, which are reprogrammable chips that can be modified post-manufacturing via software. These adaptable chips are essential for applications like AI and 5G, where standards and models are subject to frequent changes. Currently, this market is largely controlled by Xilinx and Altera, the latter of which was acquired by Intel in 2015 for $16.7 billion.

Flex Logix identified an opportunity to become “the ARM of FPGAs” by assisting other companies in designing their own chips. The company gained customer acceptance for its designs through collaborations with organizations such as Sandia National Laboratory, the Department of Defense, and Boeing. More recently, it has been developing its own chip line, the InferX X1, establishing a business model similar to what Nvidia is expected to have following the completion of its ARM acquisition and regulatory approvals.

Building on this foundation, Flex Logix has announced the availability of its X1 chip, which will be offered in four different speed grades, ranging from 533Mhz to 933Mhz. According to CEO Tate, a key advantage of the company’s offering is its pricing structure: chips will be available for between $99 and $199 for smaller quantities, and $34 to $69 per chip for large-volume orders.

Tate indicates that these chips are more cost-effective because they are considerably smaller than comparable chips from Nvidia’s Jetson series, potentially up to one-seventh the size. Smaller chips generally lead to lower manufacturing costs, as a single wafer can accommodate a greater number of chips, thereby distributing the production expenses across a larger volume. The company claims its chips deliver superior performance compared to Nvidia’s Xavier module, although independent performance evaluations are currently unavailable.

“Our customers consistently seek greater processing power for every dollar spent, and increased processing efficiency per unit of energy consumed… and our size advantage allows us to provide them with more value for their investment,” Tate stated.

Initial customer samples of these new chips are anticipated to be available in the first quarter of the coming year, with full-scale manufacturing commencing in the second quarter.

The company intends to continue operating both facets of its business, fostering growth and refining its technology. “Our FPGA business is currently profitable on a standalone basis, with revenue exceeding engineering and operational expenses. We are now also developing this new business focused on inference, which is projected to become even larger as the inference market experiences rapid expansion,” Tate explained.

The company’s board of directors includes Peter Hébert and Shahin Farshchi from Lux, Pierre Lamond from Eclipse, and Kushagra Vaid, a distinguished engineer at Microsoft Azure. Flex Logix is headquartered in Mountain View, California.