News Releases

• Opening the Door to a New Era of AI Compute
• Leading experts to explore scalable, cost-effective AI solutions beyond proprietary ecosystems

LAS VEGAS, March 11, 2025 /PRNewswire/ -- TensorWave, an AI GPU cloud platform, will launch the Beyond CUDA Summit on March 17, 2025, at the Guildhouse in San Jose, California. The event will address an industry challenge: reducing reliance on proprietary software ecosystems like compute unified device architecture (CUDA), unlocking new levels of flexibility, cost efficiency, and hardware diversity in AI development.

For over a decade, the CUDA platform has dominated AI computing, with estimates indicating that it powers between 70% and 95% of AI workloads. While CUDA has significantly advanced AI development, its dominance has led to challenges such as vendor lock-in, escalating infrastructure costs, and limited hardware flexibility. As AI compute demand continues to surge, with the global AI market projected to exceed $826 billion by 2030, the industry is actively seeking open-source and hardware-agnostic solutions to enhance performance scalability and reduce energy consumption.

"We believe that the future of AI development is heterogeneous, not monolithic," said Jeff Tatarchuk, Chief Growth Officer and co-founder of TensorWave. "By working with engineers and founders across the industry, we are building ecosystems that prioritize performance, not proprietary control."

"The future of AI isn't locked into a single ecosystem—it's open, scalable, and hardware-diverse," said Darrick Horton, CEO of TensorWave. "Beyond CUDA is a rallying point for those who want to unlock greater efficiency, cost savings, and choice in AI infrastructure."

The Beyond CUDA Summit will bring together AI engineers, researchers, and enterprise leaders to explore cutting-edge AI compute infrastructure. The summit will highlight the advantages of alternatives to CUDA, focusing on tools, software, and hardware that enable more scalable, cost-effective AI development.

Featured Speakers:

• Gregory Diamos – Co-Founder, MLCommons, providing insights into cross-platform benchmarking for fair hardware comparisons.
• Ramine Roane – Corporate Vice President of AI at AMD, discussing how model architectures are evolving for hardware diversity.
• Darrick Horton – CEO, TensorWave, who will present the latest advancements in AMD-powered AI infrastructure.
• Dylan Patel – Founder, SemiAnalysis, leading discussions on the economics of AI infrastructure and cost-saving alternatives to CUDA.

Why Attend?

• Deep Dive Presentations – Breaking down the CUDA moat and the future of pre-training, post-training, and inference.
• Networking – Connect with top AI researchers, engineers, and founders.
• Demos – Explore demos from developers and companies pushing the boundaries of what's possible with AMD.

Event Details:

• When: March 17, 2025 | 10:00 AM – 5:30 PM PT
• Where: The Guildhouse, San Jose, CA (three blocks from NVIDIA's GTC conference)
• Register: [LINK] (Free for engineers and researchers; enterprise passes available for workshops)

For additional information, please visit: www.tensorwave.com.

About TensorWave:

TensorWave is a cutting-edge AI GPU cloud platform built for high-performance AI workloads. Using AMD's MI300X accelerators and optimized software, TensorWave provides an open-source, hardware-agnostic infrastructure that competes with proprietary solutions like CUDA. TensorWave empowers researchers, engineers, and enterprises to innovate faster, with greater flexibility and reduced reliance on single-vendor ecosystems. Learn more at tensorwave.com.

Contact:
Julia Laznuik
Zen Media
+34697840113
391411@email4pr.com 

SOURCE TensorWave