Thunder Compute, a San Francisco-based startup, announced today that it has secured $13 million in Series A funding. The round was led by Matrix Partners, with participation from Y Combinator and CEAS Investments. The company aims to address the global GPU capacity shortage by virtualizing the estimated $200 billion of compute that sits idle in data centers worldwide.
The core issue Thunder Compute tackles is the remarkably low average GPU utilization rate of approximately five percent. This inefficiency stems from the fact that GPUs are typically allocated to specific tasks or workloads, and when those tasks are not running at full capacity, the hardware remains underused. Thunder Compute's proprietary virtualization software treats GPUs as network resources, allowing them to be dynamically pooled and shared across multiple workloads. This approach operates invisibly beneath existing workloads, boosting overall data center efficiency without requiring changes to applications.
“Every GPU should be a shared resource, not a dedicated one,” said Carl Peterson, co-founder and CEO of Thunder Compute, in a statement. “Our technology unlocks idle capacity, enabling enterprises to get more done with the same hardware investment.” Peterson, previously a management consultant at Bain & Company, co-founded the company in 2022 with Brian Model, a former quantitative developer at Citadel Securities.
The funding will be used to scale Thunder Compute's operations and forge partnerships with enterprises to virtualize GPUs at scale. By doing so, the company hopes to mitigate the compute shortage that has become a bottleneck for AI development and other computationally intensive tasks. The virtualization layer also offers potential benefits in cost reduction and sustainability, as more efficient use of existing hardware could reduce the need for new data center construction.
Matrix Partners, a leading venture capital firm, expressed confidence in Thunder Compute's approach. “Thunder Compute is addressing a massive market inefficiency with a technically elegant solution,” said a spokesperson for Matrix. “We believe their technology will become foundational to next-generation data center architectures.”
Thunder Compute's technology is particularly timely as demand for GPUs continues to outpace supply, driven by the explosive growth of artificial intelligence and machine learning applications. The company's solution offers a pragmatic alternative to building more facilities, which can be costly and time-consuming. By maximizing the utilization of existing GPUs, Thunder Compute helps organizations reduce their environmental footprint and optimize their infrastructure spending.
With this Series A round, Thunder Compute is well-positioned to expand its customer base and enhance its platform. The company's vision of a future where every GPU is virtualized aligns with industry trends toward greater resource abstraction and efficient scaling. As data centers increasingly adopt virtualization across all compute resources, Thunder Compute is at the forefront of this transformation.
For more information about Thunder Compute and its offerings, visit thundercompute.com.
