← All sectors / The AI transformation
082 · Enterprise search & knowledge infrastructure
Finding what the company already knows
Curve position
Binding constraint
Permissions, because a model that ignores access controls cannot be deployed.
Why this sector sits at Flight
We place enterprise search and knowledge infrastructure at Flight. The binding constraint has been whether companies will let AI search and act across their internal systems with permissions and governance intact. Glean's figures suggest many now will: about $100 million ARR in the year to January 2025, about $200 million by December 2025, and $300 million in May 2026, at a $7.2 billion valuation. Microsoft 365 Copilot is competing for the same budget, which is a Flight signal, not a Liftoff one.
We hold confidence at medium. The evidence leans on one private company, part of Glean's ARR may be usage-based, and coverage in May 2026 reported enterprise AI budgets tightening.
A move to Orbit would need growth to settle and the winners to become clear.
Reviewed on a two month cycle. The position moves only when a dated, verifiable change in the binding constraint justifies it.
Heatmap / How this sector's stocks are moving
| Stock | Price | 1 Day | 1 Week | 1 Month | Year to date |
|---|---|---|---|---|---|
| $ESTC | $89.68 | −0.87% | −15.86% | −5.20% | +18.88% |
| $MDB | $328.38 | −2.08% | −16.54% | −8.96% | −21.76% |
| $SNOW | $290.60 | −1.50% | −14.51% | −13.50% | +32.48% |
| $GOOGL | $288.93 | +3.18% | −11.07% | −16.90% | −7.69% |
| Sector average | −0.32% | −14.49% | −11.14% | +5.48% |
Prices as of November 5, 2026. The sector's move is the equal-weighted average of these listed companies, a market proxy for context, not recommendations. See it on the full heatmap →
Sentiment / The mood in this sector's stocks
Since 2018, Enterprise search & knowledge infrastructure has been in Capitulation and falling 16 separate times. After those, its stocks beat their usual three months 7 times (44%), by a median of −3.6 points.
Confidence: low, a split like this is common by chance, so no edge is called.
An enterprise model is only as useful as what it can retrieve. Most organizations have decades of documents, tickets, contracts, and messages spread across systems that were never designed to be searched together, let alone by a model.
Historically enterprise search was a poorly regarded category, delivering keyword results nobody trusted. Semantic retrieval changed the quality enough that the same buyers are looking again, with a different problem to solve.
The structural driver is that every enterprise AI deployment hits the same wall: the model is fine, the retrieval is the problem. Companies discover this after the pilot, which is why spending is shifting from models to the data layer around them.
The technology layer spans connectors to source systems, chunking and embedding pipelines, vector and hybrid search, permission aware retrieval that respects who may see what, evaluation tooling, and the caching that keeps costs manageable.
Adoption economics are justified by employee time. Knowledge workers spend a meaningful share of the week looking for information, and recovering part of that is a large number in any organization.
The beneficiaries include retrieval platform vendors, vector database companies, connector and integration specialists, and the collaboration suite vendors that already hold the content and the permission model.
The value chain runs from source systems through connectors and indexing to retrieval and the application. Whoever owns the permission model holds the strongest position, which favors incumbent suite vendors.
The overlooked layer includes integration and connector vendors, data quality and deduplication tools, evaluation and observability software for retrieval, and the services firms doing implementation.
Competitive dynamics are difficult for independents, because the largest collaboration vendors bundle retrieval into products enterprises already pay for. Specialists win where content sits outside those suites.
Risks: bundling by incumbents is the central threat, retrieval quality is hard to prove before purchase, projects stall on data hygiene, and the category has a history of disappointing buyers.
What to watch: retrieval spending disclosed separately from model spending, vector database adoption, permission aware deployments in regulated industries, and bundling announcements from suite vendors.
Coverage / Daily Disruptor issues in this sector
