#United States #Santa Clara #AI Accelerators #Velaura AI #Titan Core

Velaura AI Launches Titan Core Technology

Velaura AI, Inc., formerly known as Auradine, has introduced its groundbreaking platform, Titan Core™, aimed at transforming power efficiency for AI accelerators. The technology is set to achieve up to twice the power savings, translating to a reduction of 500W for traditional 1000W GPUs or XPUs. For hyperscale data centers, this advancement could lead to significant electrical and cooling cost reductions, potentially saving tens to hundreds of millions of dollars.

Addressing the Power Challenge

As AI workloads become increasingly intensive, power consumption has become the primary constraint in global AI-related operations. With projections indicating a doubling of electricity demand to over 945 TWh annually by 2030, ensuring efficient power usage has never been more critical. Titan Core aims to mitigate this situation by dramatically lowering the energy consumption of AI accelerators.

Velaura AI perceives the high power expenditure originating from mathematical operations, specifically matrix multiplications (MATMUL), as a primary contributor to the energy demands of AI training and inference. Its proprietary technology is reported to cut the energy required for these calculations by 2 to 4 times. This efficiency translates to savings of approximately $1,300 in electricity for each XPU over three years, coupled with reduced cooling and infrastructure expenses that can lead to a significantly lower total cost of ownership (TCO).

Seamless Integration

Titan Core integrates smoothly with existing System-on-Chip (SoC) architectures, maintaining full functional equivalence, which allows for easy implementation into customers' existing designs. Velaura AI utilizes proprietary libraries and design methodologies to create an optimized physical layout that operates on lower voltage, enhancing the overall power efficiency.

Co-founder and CEO Rajiv Khemani stated, “The future of AI will be defined by performance delivered within practical power limits. Our early collaborations with leading hyperscalers show Titan Core's potential to significantly cut global AI data center power costs.” This sentiment is echoed by industry analysts, suggesting that efficiency savings in power consumption could radically change operating metrics for data centers across the globe.

Proven Technology Stack

Velaura AI's commitment to ultra-low-power computing is supported by years of expertise in custom circuit design, covering advanced semiconductor nodes such as 3nm and 2nm. Their mature technology stack includes low-voltage digital design libraries, advanced EDA flows, and circuit methodologies optimized for low-voltage functionality, resulting in world-class yield and reliability.

The company has already shipped over 30 million efficient ASICs to establish a proven foundation for Titan Core, thus affording customers a significant time-to-volume advantage over do-it-yourself approaches that often come with high yield and reliability risks.

Significant Economic Impacts

By focusing on power efficiency, Titan Core not only promises direct economic benefits but also enables more substantial AI deployments, particularly in power-constrained environments. Other expected benefits include:

• - Lower electricity bills and operating costs.
• - Diminished cooling requirements leading to simpler thermal management.
• - Increased compute density per rack within existing power constraints.

About Velaura AI

Founded by a talented team of entrepreneurs and technology experts, Velaura AI, Inc. stands at the forefront of ultra-low power computing solutions for data centers and AI applications. Operating from Silicon Valley, California, the company merges innovative silicon design with proven intellectual property to push the boundaries of possible technological advancements.

To learn more, visit www.velaura.ai.