Ending GPU Vendor Lock-In: True CUDA Portability for the AI Era with Michael Søndergaard episode artwork

EPISODE · Jan 7, 2026 · 29 MIN

Ending GPU Vendor Lock-In: True CUDA Portability for the AI Era with Michael Søndergaard

from GSD Venture Studios Podcasts by Gary Fowler · host GSD Venture Studios

Join Michael Søndergaard, CEO and Founder of Spectral Compute, for a deep-dive conversation with Gary Fowler on one of the most critical infrastructure challenges in AI and high-performance computing: hardware lock-in within the CUDA ecosystem.Michael explains how Spectral Compute’s SCALE toolchain — a compiler and CUDA-compatible libraries — allows a superset of CUDA code to compile directly to both NVIDIA and AMD GPUs with no intrinsic performance overhead. This breakthrough enables developers and enterprises to choose the best hardware for their workloads without rewriting code or sacrificing efficiency.🎯 Key Topics Covered:✓ The problem of hardware lock-in in CUDA, HPC, and AI workloads✓ Why GPU portability is becoming a strategic advantage for AI teams✓ How SCALE delivers CPU-style portability for GPU computing✓ Supporting CUDA-X APIs like cuBLAS and cuSOLVER across vendors✓ Eliminating performance tradeoffs when moving between NVIDIA and AMD GPUs✓ What this means for cost, flexibility, and supply-chain resilience✓ Lessons from building deep tech infrastructure for developers✓ Spectral Compute’s momentum, $6M seed round, and industry recognition🌍 Why This Matters:As demand for AI acceleration surges, reliance on a single hardware vendor creates cost pressure, supply constraints, and strategic risk. Spectral Compute is redefining how developers approach GPU computing by decoupling software from hardware choice — unlocking true competition and innovation at the infrastructure layer.👤 About the Guest:• CEO & Founder of Spectral Compute• Computer scientist with over a decade of software engineering consulting experience• Creator of SCALE — a cross-platform CUDA-compatible GPU toolchain• Leading a global team breaking down GPU vendor lock-in• Secured $6M seed funding in 2025 to expand cross-platform capabilities• Featured in Business Insider for redefining GPU portability💡 Ideal For:AI engineers, HPC developers, infrastructure founders, CTOs, data scientists, cloud providers, and investors tracking the future of accelerated computing. Timely Conversation:As GPU demand outpaces supply, portability and flexibility are no longer optional — they are foundational to the next wave of AI innovation.Subscribe for more deep tech founder conversations from GSD Venture Studios: https://gsdvs.com#GPUComputing #CUDA #AIInfrastructure #HPC #DeveloperTools #SpectralCompute #MichaelSondergaard #GaryFowler #GSDVentureStudios

Join Michael Søndergaard, CEO and Founder of Spectral Compute, for a deep-dive conversation with Gary Fowler on one of the most critical infrastructure challenges in AI and high-performance computing: hardware lock-in within the CUDA ecosystem.Michael explains how Spectral Compute’s SCALE toolchain — a compiler and CUDA-compatible libraries — allows a superset of CUDA code to compile directly to both NVIDIA and AMD GPUs with no intrinsic performance overhead. This breakthrough enables developers and enterprises to choose the best hardware for their workloads without rewriting code or sacrificing efficiency.🎯 Key Topics Covered:✓ The problem of hardware lock-in in CUDA, HPC, and AI workloads✓ Why GPU portability is becoming a strategic advantage for AI teams✓ How SCALE delivers CPU-style portability for GPU computing✓ Supporting CUDA-X APIs like cuBLAS and cuSOLVER across vendors✓ Eliminating performance tradeoffs when moving between NVIDIA and AMD GPUs✓ What this means for cost, flexibility, and supply-chain resilience✓ Lessons from building deep tech infrastructure for developers✓ Spectral Compute’s momentum, $6M seed round, and industry recognition🌍 Why This Matters:As demand for AI acceleration surges, reliance on a single hardware vendor creates cost pressure, supply constraints, and strategic risk. Spectral Compute is redefining how developers approach GPU computing by decoupling software from hardware choice — unlocking true competition and innovation at the infrastructure layer.👤 About the Guest:• CEO & Founder of Spectral Compute• Computer scientist with over a decade of software engineering consulting experience• Creator of SCALE — a cross-platform CUDA-compatible GPU toolchain• Leading a global team breaking down GPU vendor lock-in• Secured $6M seed funding in 2025 to expand cross-platform capabilities• Featured in Business Insider for redefining GPU portability💡 Ideal For:AI engineers, HPC developers, infrastructure founders, CTOs, data scientists, cloud providers, and investors tracking the future of accelerated computing. Timely Conversation:As GPU demand outpaces supply, portability and flexibility are no longer optional — they are foundational to the next wave of AI innovation.Subscribe for more deep tech founder conversations from GSD Venture Studios: https://gsdvs.com#GPUComputing #CUDA #AIInfrastructure #HPC #DeveloperTools #SpectralCompute #MichaelSondergaard #GaryFowler #GSDVentureStudios

NOW PLAYING

Ending GPU Vendor Lock-In: True CUDA Portability for the AI Era with Michael Søndergaard

0:00 29:50

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

No similar episodes found.

No similar podcasts found.

Frequently Asked Questions

How long is this episode of GSD Venture Studios Podcasts by Gary Fowler?

This episode is 29 minutes long.

When was this GSD Venture Studios Podcasts by Gary Fowler episode published?

This episode was published on January 7, 2026.

What is this episode about?

Join Michael Søndergaard, CEO and Founder of Spectral Compute, for a deep-dive conversation with Gary Fowler on one of the most critical infrastructure challenges in AI and high-performance computing: hardware lock-in within the CUDA...

Can I download this GSD Venture Studios Podcasts by Gary Fowler episode?

Yes, you can download this episode by clicking the download button on the episode player, or subscribe to the podcast in your preferred podcast app for automatic downloads.
URL copied to clipboard!