Here is a companion to The Intelligence Foundry research and analysis.
Our previous market assessment in this series named the private Applied-AI Initiatives for elite professional services — Kirkland & Ellis and Palantir's roughly $500 million fund-formation engagement, Harvey, Intapp, and Microsoft's emerging context-layer stack.
The natural next question is how large this market opportunity actually is, for comparison.
The honest answer is more useful than a confident one: no tech industry analyst firm currently tracks this category directly, and the closest proxies disagree with each other about the upside growth potential by two to four times.
Here's what can be said with real numbers behind this estimate.
"Firm-owned AI" — a governed knowledge graph plus a fine-tuned model trained on an organization's own proprietary judgment, as distinct from a frontier model consumed off the shelf — is not a line item any tech market research firm reports on its own.
It sits inside broader categories (AI software, AI platforms, legal technology, sovereign AI) without being cleanly separable from them. Any single dollar figure claiming to size it precisely is manufacturing false precision. What follows instead is the real data on the categories this opportunity lives inside, and a transparent, assumption-shown estimate of the slice that belongs to it.
Three data points, each from a named source, bound the broader market this sits within:
IDC forecasts the worldwide AI software market — the total addressable pool, generic and proprietary spend combined — will grow from $64 billion in 2022 to nearly $251 billion by 2027, a 31.4 percent compound annual growth rate.
Gartner's narrower "AI platforms and models" segment, arguably the closest tracked category to a firm-owned-AI stack, is forecast to reach $64 billion in 2026, up 63.4 percent from $39 billion in 2025 — a segment growing roughly twice as fast as AI software overall.
IDC separately projects total enterprise AI solution spend, software and services combined, at $307 billion in 2025, rising to $632 billion by 2028.
These are ceilings, not answers. Firm-owned AI is a slice of each of them, not the whole.
Two adjacent categories have actual analyst coverage, and both show how immature this space still is in reality.
Sovereign AI — the nearest regulatory-driven analog, though it's mostly about national data-localization infrastructure rather than firm-level proprietary knowledge — is sized at $40 billion in 2025 growing to $148 billion by 2032 by one research firm, and at $15 billion in 2025 growing to $177 billion by 2035 by another. McKinsey, separately, puts the entire sovereign AI opportunity at roughly $600 billion by 2030. Three credible sources, three different starting points, none reconciling cleanly.
Legal AI software — the vertical the Kirkland and Harvey case studies actually sit in — shows the same pattern at smaller scale: one estimate has it at $3.11 billion in 2025 growing to $10.82 billion by 2030; another puts 2024 at $1.45 billion growing to only $3.90 billion by 2030. Same nominal market, same year, roughly a 2 times gap.
When established research firms can't agree within a factor of two on a category as concrete as "legal AI software," a coined term like "firm-owned AI" deserves a range, not a precise point estimate.
Working from the anchors above: if the U.S. represents something like 40 percent of global AI software spend — a plausible but not independently confirmed share, offered as an assumption rather than a sourced fact — that puts the U.S. AI software market in the neighborhood of $100 billion by 2027.
Firm-owned, proprietary-knowledge-driven AI is a fraction of that total, not a majority of it; using the growth rates already visible in legal AI and sovereign AI cloud as a bellwether for how fast the proprietary slice is expanding relative to generic AI spend.
A reasonable order-of-magnitude read is: low single-digit billions in U.S. firm-owned AI investment today, scaling toward roughly $15–30 billion by 2030 as adoption moves past legal into consulting and corporate finance.
That range should be read as a synthesis of public market data, not a forecast anyone has published. Meaning, anyone using it in a client conversation should say so.
A market-sizing number is not the reason to act in this category — the case studies in our prior assessment already make that case on their own terms. Three practical conclusions follow instead:
Don't let a manufactured TAM anchor a budget conversation. The categories with real analyst coverage disagree by 2–4 times; a business case built on a single precise-looking number borrowed from an adjacent, ill-fitting category won't survive scrutiny from a sharp CFO.
Watch the vertical-specific signals instead. Legal AI's growth rate, however inconsistently sized, is the most mature paper trail this category has to date. Consulting and corporate finance lag — which is where the white space opportunity lies, not where the market-sizing certainty is at this time.
Treat every figure here, and every figure in any AI vendor's pitch deck, the way this page treats its own: state the source, state what it actually measures, and flag anything self-reported as exactly that.
This estimate draws on published forecasts from IDC, Gartner, and McKinsey, and market-sizing reports from MarketsandMarkets, Grand View Research, and Precedence Research, current as of August 2026.
Where sources disagreed, both figures are shown rather than one being silently chosen. The U.S. share-of-market and proprietary-segment assumptions used to build the working estimate are my own synthesis, not a published tech industry analyst figure, and should be treated accordingly.
Interested in what an Intelligence Foundry strategy would mean for your organization?
Learn more > Contact us now