Executive Summary
AI investment reached $47 billion in H1 2026, up 34% QoQ but down 12% YoY as investors pivot from model-layer speculation to infrastructure and deployment capabilities. The dominant narrative has shifted from “train bigger models” to “deploy cheaper at scale.”
Investment Distribution by Layer
| Layer | H1 2025 | H1 2026 | Change |
|---|---|---|---|
| Foundation Models | 42% | 28% | -14pp |
| Infrastructure | 23% | 35% | +12pp |
| Orchestration | 18% | 22% | +4pp |
| Applications | 17% | 15% | -2pp |
Infrastructure — particularly GPU supply chain, vector databases, and observability tools — captured record inflows despite overall contraction vs. 2025 peak.
Notable Rounds (May-July 2026)
- Fireworks AI Series C: $200M @ $1.2B valuation (serving inference at scale)
- Lakera Series B: $85M @ $450M valuation (model security and testing)
- Pulumi AI Series A: $50M @ $280M valuation (infrastructure orchestration)
- Weights & Biases late-stage: $175M @ $1.5B valuation (MLOps platform)
Strategic Implications
Investors now prioritize revenue visibility over growth-at-all-costs metrics. Companies demonstrating path to profitability within 18 months command premium valuations; pure experimentation funding dried up mid-year.
Contact for licensing: [email protected]