The Agentic Ecosystem Index
First, you missed NVIDIA. Then you missed memory - SK Hynix, Micron, commodity cyclicals that got repriced as structural growth stories once Blackwell made HBM the critical constraint.
Something bigger is forming now, the agentic economy. AI agents are crossing the viability threshold. Not only because models got smarter, but because a wave of software optimizations is about to cut inference costs by 10-20x (not only TurboQuant). A $3-5 AI contract review (cost of today) becomes $0.30 (in 12 months). At that price, you run it on everything.
But the first viable agents won’t replace cheap repetitive tasks. They’ll replace expensive human judgment — $500/hour legal reviews, M&A due diligence, clinical workflows. Because only high-value tasks absorb messy inference costs today. The cheap stuff comes later. This creates a massive investable ecosystem.
We built the Rubin Build-Out 100 to track the supply side of AI — 18 sectors, 100 companies forming the physical infrastructure constraint that NVIDIA cannot route around. It has worked exceptionally well for investment purposes. Now the demand side is forming.
We’re building the Closelook Agentic Ecosystem Index to track the companies that turn tokens into economic output.
The structure: 7 layers, 24 sectors, 130-150 companies with global exposure.
Layer 1 — Silicon & Architecture. GPU design, edge inference chips, chip IP, semiconductor equipment. Rubin overlap, but weighted for inference volume rather than node advancement.
Layer 2 — Inference Infrastructure. Neoclouds, hyperscale platforms, optimization software. Where cost deflation happens.
Layer 3 — Toll Booth Infrastructure. Observability, cybersecurity, data infrastructure, edge delivery, identity, communication APIs, and power. Companies that win on rising transaction volume, regardless of which application prevails.
Layer 4 — Foundation Models & Orchestration—the companies producing tokens and the frameworks coordinating agents—heavily private today.
Layer 5 — Horizontal Agent Applications. Coding agents, enterprise workflow automation, search. Where some beaten-down SaaS incumbents get repriced.
Layer 6 — Vertical Agent Applications. Legal, healthcare, financial services, and industrial operations. Highest margins, deepest moats, 3-5x higher retention than horizontal solutions.
Layer 7 — Backdoor Exposure. Public companies owning stakes in private agentic companies.
We’ll add a viability timeline overlay. Each sector tagged Phase 1 (viable now), Phase 2 (viable once optimization hits, 2027-2029), Phase 3 (viable at commodity inference costs, 2029+). The thesis shifts as costs fall.
And a two-tier structure: ~90 public companies, plus a ~50-company private pipeline with IPO readiness, and backdoor public exposure.
The Rubin tracks what enables inference. The Agentic Index tracks what inference enables. Together, they map the AI economy.
Full breakdowns and company maps will soon be published on closelook.net.