Industry
Consolidation Will Not Be the One You Expect
We imagine consolidation through acquisitions until three survivors remain. In legal AI, it will happen between layers.
When one speaks of consolidation in a market that has widened too much, one generally imagines the same scenario: a few players absorb the others, the market shrinks to three or four main vendors, and the story closes. This image is useful in certain sectors, where it faithfully describes what happens. It is probably misleading for legal AI, because the consolidation coming in this market will not be a consolidation between equivalent players buying each other out. It will be a consolidation between layers.
Two images of consolidation
One must explain what this distinction covers, for it entirely changes the reading of the market. A consolidation between players reduces the number of vendors: the big eat the small until only a few remain. A consolidation between layers does not reduce the number of players, it reorganizes them: the tools stay numerous, but a layer installs itself above them and holds them together, so that the question ceases to be “who will survive” and becomes “who will cap the others.” It is not a reduction, it is a putting-in-order.
You expect a consolidation between players. It will be a consolidation between layers.
This distinction is not only descriptive, it is predictive. If consolidation happens between layers and not between players, then the number of tool vendors will not necessarily decrease in the coming years; it could even keep rising, while the layer that organizes them installs itself above. Those who watch for signs of consolidation by counting buyouts and bankruptcies will look in the wrong place: the real consolidation will not be read in the reduction of the number of players, but in the emergence of a layer that more and more organizations adopt to cap their stack.
The reason one gets the image wrong owes to a habit of thought. We have seen so many markets consolidate by buyout, from telecoms to retail, that we apply this model by default to any market with too many players. But this model supposes the players are substitutable, that absorbing one amounts to roughly absorbing another. In legal AI, the tools are not substitutable: each does well one thing the others do poorly, and forcibly uniting them by merger would destroy precisely the specialization that makes their value. It is this non-substitutability that makes buyout consolidation ill-suited, and layer consolidation natural.
Why no single player suffices
The legal AI market today counts several dozen vendors, and this number keeps rising. But as mature organizations formulate their needs more precisely, one realizes that none of these vendors, taken alone, matches what buyers demand. Some are excellent drafting tools, but without operational memory. Others have good research, but without governance. Some do limited orchestration, but without integration to existing tools. None carries the complete semantic layer a serious organization expects.
This situation can resolve in two ways, and it is the choice between them that decides the market’s form. The first: acquisitions and mergers reconstitute, inside a single player, the set of missing functions. It is classic consolidation, possible but slow, costly, and supposing a player has at once the craft depth, the architecture and the absorption capacity needed. The second, more likely in the short term: a unified semantic layer installs itself above the stack of existing tools and makes them coherent. The underlying tools do not disappear, they become components of an upper layer. Consolidation happens by architectural overlay, not by commercial absorption.
One can understand why the first scenario, that of mergers, is so slow to fall into place. Buying a player does not suffice to integrate its product: one must merge architectures conceived separately, reconcile different data models, harmonize divergent usage logics, all without breaking what worked. These integrations take years and often half-fail. Meanwhile, the unifying layer does not need to merge the tools: it lets them be what they are and merely coordinates them from above. Where the merger undoes to redo, the layer composes without undoing.
Piecing players together by merger is slow and costly. Capping them with a layer is fast and reversible.
Reversibility deserves a pause, for it is a decisive advantage of the layer over the merger. A merger is irreversible: once two products are integrated, one no longer separates them without breaking everything, and the client organization is bound to the result, good or bad. A layer, on the contrary, leaves the tools replaceable beneath it: if one of them ceases to suit, one changes it without touching the layer or the others. Layer consolidation therefore preserves the organization’s freedom where buyout consolidation reduces it, and this preserved freedom is one more reason the market will lean toward the layer rather than the merger.
This second scenario is structurally faster and more efficient. It avoids the integration costs of mergers, preserves the specialization of the underlying tools, lets organizations keep what they have already deployed. And above all, it concentrates strategic value not in the biggest vendor, but in the player who will have built the unifying layer. It is a displacement of value from volume to architecture: it is no longer size that decides, but the ability to hold the others together.
One must measure how unusual this displacement is. In a buyout consolidation, the winner is the biggest, the one with the means to absorb the others; the game plays out at the balance-sheet level. In a layer consolidation, the winner is the architecturally rightest, the one who built the coordination point everyone ends up adopting; the game plays out at the design level. A small player well positioned architecturally can thus prevail over large poorly positioned vendors, which is unthinkable in an absorption consolidation.
It is this dynamic of layer consolidation that MAX anticipates and accompanies. The Legal Semantic Layer is not meant to replace existing tools; it is designed to coordinate them, enrich them, make them coherent within an organization. This posture corresponds precisely to the form the real consolidation of this market will take. Organizations waiting for a consolidation by acquisition to deliver them a ready-to-use solution will wait a long time, probably in vain; those that choose the unifying layer now build, without always knowing it, the final form of the market.
The market will not consolidate around a tool. It will consolidate around the layer that makes the tools coherent.
