While artificial intelligence burns energy and water at an increasing rate, a Tampa-based AI startup is working with one of the world's biggest technology companies on efficient AI.
Startup Lumina AI joined Intel's Liftoff program, a virtual program for early-stage AI startups that aims to help companies scale and give access to resources. The move toward the market focuses on Lumina's product and is a way to validate the startup's goal, said Daniella Diaz, vice president of revenue and marketing.
"For Lumina, it is our next phase of growth," Diaz said. "It is our ability to continue to meaningfully deliver our algorithm at scale with the help of Intel and their technical teams and expertise."
With acceptance to the program, Lumina will focus on improving its recently unveiled "random contrast learning" algorithm, a text classification process based on central processing units. In late 2023 and early 2024, Lumina finished developing the algorithm and finalized its white paper on the proprietary concept. The startup has been in Tampa since its founding in 2015.
A central processing unit, the computer chip that Intel makes and is generally considered the computer's brain, has been overtaken in the tech world through a focus on the graphics processing unit. California's Nvidia is known for its GPUs, which handle video game visuals or complex computations. GPUs exploded in the tech space because GPU-powered devices are advantageous for AI.
The problem with focusing on GPU-enabled AI computing is energy consumption. The devices require massive amounts of energy, and consumption is becoming a concern for electric and water sustainability. Lumina AI seeks to accomplish this with its new specialized algorithm that uses CPUs: a more efficient and less consumptive model for AI computing.
With Intel, Lumina plans to tap into its CPU expertise and further this method of AI computing, Diaz said.
"It's competitive, expensive and cost prohibitive for both individuals and organizations to even start thinking about what their AI partnerships are going to look like, how to utilize AI in their businesses," Diaz said. "We hope that our algorithm can continue to democratize this machine learning technology."
