AI Chip Startup Oxmiq Raises $35 Million to Cut the Cost of Artificial Intelligence Computing

AI Semiconductor startup Oxmiq raises $35M to make AI chips cheaper and more efficient. The company is developing a novel chip architecture to dramatically reduce the cost and power consumption associated with artificial intelligence compute. This funding will help the startup bring its technology to market by supporting product development, engineering hires and commercialization.

"Our new processor will make training and running these sophisticated models significantly more affordable and energy-efficient," Oxmiq CEO and founder Steven Verleye said.

"This investment is critical for enabling us to move from concept to silicon and offer our transformative AI computing solution to the market." The investment rounds comes amid explosive growth in AI demand, particularly for the training and inferencing of modern language models, where enterprises and researchers are shelling out billions for graphics processing units (GPUs) and other AI accelerators. The cost of computing has emerged as one of the biggest hurdles for adopting and scaling artificial intelligence technologies. Organizations already spend tens or even hundreds of millions dollars annually for compute power, creating a massive incentive for startups like Oxmiq that can bring down the expense of AI infrastructure.

Instead of building chips that just push for greater raw performance in the same way as conventional chips, Oxmiq’s designers are rethinking the basic architecture and processes of how AI algorithms operate, hoping to run the same operations with fewer steps and less electricity.

They aim to hit a balance that still affords the performance that models need for increasingly complex machine learning tasks. Making AI more cost-effective has become the industry’s foremost challenge. "It's become more important than ever to make computing accessible to run this rapidly scaling technology," Verleye added.

"Our chip is about 20x more efficient than standard hardware." This investment will enable Oxmiq to continue building out its team, accelerate development and bring the company's first generation of processors closer to commercialization, at a time when competition for the future of AI hardware heats up. Other startups in the market aim to compete for the burgeoning opportunity of silicon that specifically addresses AI workloads, including inference, training, and other tasks.

The broader semiconductor landscape, meanwhile, is also evolving at an unprecedented rate, as tech companies pour billions into custom chip designs, which serve a diverse set of needs for data centers, consumer devices and more.

"There is an explosion in compute requirements, driven by generative AI, and there has never been a better time to create truly innovative compute architectures," said Ventureast Partner Ashu Lunia, one of the new investors in Oxmiq. The challenge of the tremendous amount of energy that AI centers gobble up is also in full view. This has become a crucial concern, given the implications for the operating costs, capacity and carbon footprint of global IT infrastructure. More efficient chip architecture can also offer a path to increasing sustainability in cloud computing and at the enterprise level.

More cost-effective AI hardware can also bring sophisticated AI to more applications and industries, from health care and financial services to manufacturing, education, and research.