The AI Rebuttal That Has to Survive the Evidence
This series examines the U.S. Department of Justice’s review of the proposed merger between Seismic Software and Highspot, two companies that sell sales enablement software to businesses. The central question is narrower, and more useful, than whether artificial intelligence is “disruptive”: when can the arrival of AI-native competitors actually answer a concern that a merger would remove meaningful competition?
The question matters because merger review looks forward. It does not wait for a combined company to raise prices, narrow choice, or stop innovating. It asks what competition the transaction may remove and whether the remaining market will still impose a credible constraint. A claim about future technology can therefore be relevant. But relevance is not proof. A new technology must become competition in the market that matters, for the customers who matter, on a timeline that matters.
The case is not “AI versus antitrust”
Seismic and Highspot announced a definitive agreement to merge on Feb. 12. Both offered sales enablement software platforms to businesses. The Antitrust Division opened an investigation and issued Second Requests, requiring additional production from the merging companies. The companies were not distant firms hoping to enter one another’s field; they were established rivals in the same broad software space.
The Division’s public statement says that it later made a decision to close the investigation after a targeted review of key competitive questions. The resolution reduced the companies’ Second Request compliance burden, and the Division described the matter as an example of focusing an expedited review on dispositive issues. That is an important outcome, but it is not the same thing as a general exemption for mergers involving AI. The statement does not say that AI automatically justifies consolidation, that every AI-native firm will defeat a merger concern, or that a court has declared this type of transaction lawful.
The more revealing question is what the Division had to test. The merging companies and other legacy providers had competed in sales enablement software without meaningful entry from larger, diversified technology companies operating in adjacent markets. That history could support a concern that established rivals were not easily replaceable. At the same time, newer AI-native firms were reportedly growing quickly to win sales enablement customers and increase competitive pressure on those legacy providers.
That creates two different kinds of competition. Seismic and Highspot represented rivalry that already existed when the merger was proposed. AI-native firms represented a possible source of future pressure. The second proposition could weaken the first concern, but only if future pressure became sufficiently real, fast, durable, and comparable to the competition the merger might remove. “There will be more innovation” is not yet an answer to “who will constrain these two firms’ customers after they combine?”
The distinction also explains why the history of entry matters. If larger technology companies have repeatedly possessed adjacent capabilities but have not meaningfully entered, that absence can be evidence that the market is harder to penetrate than a technology narrative suggests. It does not prove that AI-native firms cannot enter; it makes the Division’s question more specific. What changed? Did AI lower a practical barrier, change the cost or speed of building the product, create a new route to customers, or simply create a more persuasive story about a barrier that remains? The answer has to be tied to sales enablement customers rather than to the general pace of AI development.
The public statement also leaves open an important distinction between pressure and replacement. An AI-native provider might force legacy firms to improve while still being too small, too narrow, or too unstable to replace the competitive role of one of the merging companies. Conversely, a firm can matter before it matches an incumbent’s revenue if customers view it as a credible alternative and incumbents change their conduct in response. The question is therefore not whether the entrant is already as large as Seismic or Highspot. It is whether its competitive effect is durable and substantial enough to prevent the merger from materially changing the market.
The question hidden inside the closure
The Division explicitly addressed a familiar argument from merging companies: AI is a disruptive force and a reason to consolidate close competitors in industries with historically high barriers to entry. Its stated response was to evaluate that argument case by case, using ordinary-course documents and data from the merging companies and third parties.
That emphasis changes how the story should be read. The important evidence is not merely a technical description of what an AI system might eventually do. It is evidence connecting the technology to competitive behavior: customers choosing an AI-native product, incumbents reacting to that choice, product capabilities reaching a usable level, or an entrant building the scale and durability needed to remain a constraint. The public statement does not identify the firms involved, publish customer counts, or disclose the underlying win-loss data. It does, however, identify the kind of market fact that can matter: AI-native firms were winning customers and increasing pressure on legacy providers.
The procedural response was similarly focused. The Division and the merging parties entered a timing agreement that prioritized the production of evidence relevant to AI entry and repositioning. Over the following three months, the Division reviewed documents, analyzed data, and interviewed industry participants. In other words, the question was not left at the level of a prediction about technological change. The review was organized around whether entry or repositioning could change the competitive conditions in sales enablement software.
That focus is significant even though the release does not publish the record. A targeted review can ask whether the alleged entrant has a workable path to customers, whether its growth is more than a short-lived spike, and whether it would exert pressure soon enough to matter. It can also test the opposite possibility: that the merger would make entry harder, or that the same conditions said to make AI entry easy would not actually produce a substitute with comparable reach. Those are different factual questions, and a short public statement cannot answer all of them. It can show what the Division treated as dispositive without exposing every document behind the decision.
The result should therefore be read as a calibrated administrative judgment. The Division said that multiple types of evidence indicated growing AI-native pressure and that, on the particular facts, the investigation could close. It did not say that a projected entrant would always be credited, that the market had permanently changed, or that a merger can manufacture its own defense by making the post-merger market more profitable for newcomers. The difference between those claims is the difference between a case study and a safe harbor.
This is the first useful mental model for the series. The merger concern starts with current rivalry. The AI claim introduces a possible future constraint. Evidence must connect the two. If the connection is weak, the claim remains a forecast. If it is strong, the claim can become part of the reason the competitive risk is less likely than it first appeared.
Why another merger case matters
The Tapestry–Capri litigation provides a useful contrast—not because luxury handbags and sales software are the same market, but because the court’s reasoning makes the boundary problem visible. The Federal Trade Commission sought to block Tapestry’s proposed acquisition of Capri, arguing that the transaction would substantially lessen competition in an accessible-luxury handbag market. The accessible court record describes market definition as a necessary predicate to assessing competitive effects and treats the 2023 Merger Guidelines as persuasive rather than binding. The docket record reports that the court granted the FTC’s motion for a preliminary injunction.
The comparison shows why a broad story about future competition is never enough by itself. In the handbag case, the relevant questions included what products customers could reasonably switch between, how brands described their own product lines, and whether data choices distorted the apparent boundaries of the market. Docket-level summaries also point to evidence about wholesale sales and the generality of Tapestry’s third-party manufacturing relationships. Those details matter because a market is not defined by the most expansive description of an industry. It is defined by the competitive constraints that customers and suppliers actually experience.
The procedural posture matters too. The Tapestry court was deciding whether to grant preliminary relief while the FTC’s administrative case continued, not issuing a universal rule for every luxury brand or every merger involving a changing market. The accessible opinion explains that a plaintiff must propose a proper market and show likely anticompetitive effects, while a defendant can offer evidence that the prediction is inaccurate or the supporting evidence is unreliable. That structure is useful for this series because it keeps the comparison tied to proof: a court may consider future conditions, but it still needs a defined competitive arena and evidence about how firms constrain one another.
The same discipline applies to AI. An AI-native firm may be technologically impressive and still fail to constrain Seismic or Highspot if it cannot reach the same customers, solve the same purchasing problem, or sustain a comparable offering. Conversely, a smaller entrant may matter greatly if its customer wins force legacy firms to improve, lower prices, or invest faster. The label “AI-native” does not settle either possibility. Evidence does.
The next step is therefore to define the merger concern before judging the disruption claim. What competition would the transaction remove? What market contains that competition? And what would a future entrant have to do—not just promise—to replace it?