The Deal Structure That Avoided Antitrust Is Now on Trial

For several years the technology industry has used a transaction structure that achieves nearly everything an acquisition achieves while avoiding most of what an acquisition triggers. A large company hires the founders and much of the engineering team of a smaller one, licenses its technology for a substantial sum, and leaves the original entity technically alive — funded, staffed by whoever remains, and free to continue serving its existing customers.

No company is acquired. No merger notification is filed. Investors receive a return through the licensing payment. The acquirer gets the team and the technology.

That structure is now being examined by competition authorities, and the outcome will shape how consolidation happens for the rest of the decade.

Why the structure was invented

Merger review is slow, uncertain and public. For a large technology company, a straightforward acquisition of a smaller competitor in an adjacent market can mean a year or more of review, a detailed disclosure of internal strategy, and a meaningful probability of prohibition or conditions.

The licence-and-hire structure sidesteps most of that. There is no change of control, so notification thresholds may not be met. There is no acquisition of a competitor, at least not formally. The technology transfers by licence rather than by ownership, and the people transfer as individual employment decisions, which competition law generally does not regulate.

It is elegant, and for several years it worked, which is why it spread rapidly through the AI sector, where the scarce assets are people and model weights rather than factories or customer contracts.

What regulators appear to be testing

The scrutiny now underway asks whether substance should prevail over form. Edgewisely’s account of the antitrust probe putting the industry’s favourite consolidation structure on trial frames the question the way an enforcer would: if the effect of a transaction is that an independent competitor ceases to be a meaningful competitive constraint, does the legal form of the transaction matter?

The counterargument is serious and should not be dismissed. The company still exists. It still serves customers. Its investors were paid. Employees moved voluntarily, and non-compete restrictions on engineers are unenforceable in several major jurisdictions. Treating a licensing agreement plus a hiring wave as a merger would require extending doctrine considerably, and the boundary is genuinely hard to draw: at what point does hiring talented people from a competitor become a reportable transaction?

The stakes are high on both sides. If the structure survives scrutiny, it becomes the default route for consolidating any technology company whose value is concentrated in people. If it does not, a wave of completed transactions may face retrospective examination, and the market for small AI companies changes materially.

The adjacent structure: buying distribution

There is a related pattern worth reading alongside this one, and it points to where value is genuinely accumulating.

Several notable acquisitions in the last year have been justified publicly on technology grounds while the actual asset acquired was distribution — a channel, a set of enterprise relationships, an installed base that would take a decade to build. Edgewisely’s analysis of an acquisition where the messaging network, not the AI, was the real logic makes that case directly.

This is rational in a market where model capability is converging. If several vendors can deliver comparable technical results, the binding constraint on revenue is access to customers, and access to customers is expensive, slow to build and not commoditising. Buying a channel is buying the scarce thing.

It also has a competition dimension that receives less attention than model-layer consolidation. Concentration in distribution channels can foreclose competitors more effectively than concentration in technology, because a superior product with no route to the buyer is not a competitive constraint in any practical sense.

Financing as a third path

The third structure in wide use avoids acquisition entirely: invest in a partner or customer on terms that tie them to your standards.

A convertible investment that locks a chipmaker into a proprietary interconnect is not a merger by any definition, and it achieves a durable strategic outcome — Edgewisely’s reporting on a $3.5 billion investment structured to secure standards adoption shows how effective this can be. The investee remains independent, competes in its own market, and builds its roadmap around your architecture.

For competition authorities this is harder still. Minority investments generally fall below notification thresholds, standards adoption is a commercial decision, and the competitive effect is diffuse and slow. Yet the cumulative effect of a dominant firm financing the architecture choices across its supply chain is a market shaped to its advantage without a single reportable transaction.

What it looks like from the founder’s side

The commentary on these structures is almost entirely about the acquirers and the regulators. The people with the least agency in them are the founders and employees of the smaller company, and their position deserves more attention than it gets.

For a founder, the licence-and-hire outcome is frequently the best available. The alternative is often an unfundable company in a capital-intensive market against competitors with effectively unlimited resources. A structure that pays investors, secures the team’s employment on strong terms and places the technology somewhere it will actually be used is a considerably better outcome than a slow decline.

For employees who do not receive an offer, it is a worse outcome than an acquisition. In a conventional deal the whole company transfers, and retention packages usually cover the team broadly. In this structure a subset is selected, and those left behind hold equity in an entity whose founders and best engineers have departed, with a licensing payment that flowed mostly to preferred shareholders.

That distributional detail is one of the stronger arguments for treating these transactions as what they functionally are. It is also a reason to be careful about prohibiting them outright: a rule that removes this exit without providing another one does not save the company, it simply removes the option that produced any return at all.

Whatever doctrine emerges should probably distinguish between structures used to escape review of a genuinely competitive overlap and structures used to rescue a company with no independent future. Those are different transactions wearing the same clothes.

What dealmakers should take from this

Three practical implications.

Structure no longer guarantees immunity. Transactions designed principally to avoid review are now more likely to attract it, and a deal rationale that only makes sense as regulatory avoidance reads badly in discovery.

Document the commercial logic. Transactions with a clear, contemporaneous business rationale beyond removing a competitor survive scrutiny far better than ones where the internal record is thin.

Assume retrospective review is possible. Enforcement is moving toward examining completed transactions. Deals closed on the assumption of permanent finality are carrying a risk that was not priced.

The broader point is that competition law adapts to structure with a lag, and we are in the lag. The structures being used now were designed against the previous generation of enforcement. The next generation is being written in response to them, and it will be applied to deals signed today.