The demand for AI compute is exploding, creating a severe and worsening global shortage of GPUs and data center capacity that is now the primary bottleneck for industry growth.
AI-native companies are experiencing unprecedented hyper-growth (e.g., 6-10x YoY for Together.ai, Anthropic doubling revenue in 3 months), signaling a fundamental economic shift that defies traditional software valuation metrics.
The AI market has reached an inflection point where inference workloads now exceed 50% of all GPU usage, marking a transition from an R&D phase to a widespread, revenue-generating deployment of AI services.
AI model capabilities are advancing at a breakneck pace, particularly in software engineering, where performance on key benchmarks has jumped from 1% to 76% in two years, effectively industrializing the field of coding.
12 quotes
Concerns Raised
The severe and worsening shortage of GPU and data center capacity is the main bottleneck for industry growth.
The business growth of even the largest AI companies is ultimately constrained by available compute.
Public market investors may misunderstand the new economics of AI, incorrectly punishing companies for necessary infrastructure investments.
Opportunities Identified
Building infrastructure platforms like Together.ai to serve the massive demand for efficient AI inference and training.
Developing AI-native applications that can achieve hyper-growth by leveraging increasingly capable open-source models.
The emergence of new AI modalities like voice and physical AI (robotics) will create new waves of demand for compute.
Investing in the AI supply chain as the demand for inference-optimized infrastructure continues to outpace supply.