The Economics of Open-Weight Inference
How open-weight demand keeps older NVIDIA GPUs earning for years.
- GPU economics
- Compute markets
AI compute research and GPU market publications
Ornn publishes first-party research on GPU rental prices, compute markets, depreciation, procurement, memory, open-weight inference, and compute futures.
On Sovereign Compute
In the world of compute, a lot has changed in the last twelve months.
- Sovereign compute
- Compute markets
- Derivatives
The Basis Gap in Compute Acquisition Structures
Why Basis Risk Defines Every Procurement Decision
- Procurement
- GPU economics
- Infrastructure finance
Compute in Space. Why, How, and When?
The emergence of AI has led to tremendous demand for compute.
- Compute markets
- Infrastructure finance
- GPU economics
On GPU Depreciation
The last couple of months there has been strong commentary from investors — many of whom we admire — surrounding the current depreciation schedules of GPUs.
- GPU economics
- Infrastructure finance
Memory: How It Works and Why It Sets the Cost of Compute
The AI systems of today are incredibly powerful - they can write code better than most software engineers, analyze documents, generate images, and can reason through complex problems to reach coherent solutions.
- Memory
- GPU economics
- Compute markets
The AI Chip Wars: From Supplier Dominance to Contested Equilibrium
Recent reporting that OpenAI has begun running a portion of its training workloads on Amazon’s Trainium has been widely interpreted as a challenge to NVIDIA’s technical leadership.
- Compute markets
- GPU economics
- Procurement
Circular Financing in AI
How Compute is Pulled Forward - and Risk is Reallocated
- Infrastructure finance
- Compute markets
Compute as a commodity isn't oil. It's electricity.
How to hedge compute - Part 1
- Derivatives
- Compute markets