What a Merger Concern Actually Claims
For Seismic and Highspot, the question is not simply whether AI-native firms may become important. Before that question can matter, one must identify what competition the proposed merger could remove. This chapter builds that baseline: how merger law describes a threatened loss of competition, why market definition comes first, and what kind of evidence can challenge the initial prediction without turning general information into legal advice.
Competition is a set of constraints, not just a list of rivals
People often describe competition by counting firms. That is useful, but incomplete. A firm constrains another when its presence changes what the other can profitably do: how much it charges, how quickly it improves a product, which customers it pursues, how much service it provides, or how much it invests in quality. A rival that never wins a customer may still matter if the incumbent must keep its offer attractive because customers could switch. Conversely, a nominal rival may exert little pressure if its product serves a different need or cannot reach the same buyers.
The 2023 Merger Guidelines describe competition in these terms. Competition is a process of rivalry that can produce lower prices, better quality, more innovation, greater resilience, and more choice. A merger can lessen competition when it diminishes those constraints, reduces the number or attractiveness of alternatives, or reduces the intensity with which market participants compete. The point is forward-looking: the agencies examine the risk created by the transaction rather than waiting for the combined firm to demonstrate the harm after closing.
In enterprise software, the relevant constraint may not appear first as a posted price. It may appear in the speed of product releases, the willingness to customize a platform, the quality of implementation support, the terms offered to a large buyer, or the resources devoted to integrating a new capability. That does not make the analysis less concrete. It means the evidence must identify the dimension on which the firms actually compete. A merger that leaves many nominal vendors in the directory can still matter if only a small number can credibly serve the same customers and solve the same operational problem.
That matters in sales enablement software because Seismic and Highspot were described as current providers in the same broad space. The public DOJ statement does not provide a complete economic model of their rivalry, and it does not disclose the internal documents or customer-level data behind the review. It does establish the setting: both firms offered sales enablement platforms, the Division issued Second Requests, and the investigation considered whether the merger might harm competition in that market.
The first analytical task is therefore to identify the constraint that would disappear if the firms combined. Would customers lose an alternative with different functionality? Would the firms stop responding to each other’s product improvements or sales efforts? Would a buyer face fewer credible choices when negotiating terms? The precise answer is not supplied in the public closure statement, so it should not be invented. But the questions show why “AI is changing software” cannot be the starting point. The starting point is the competition at risk.
A future entrant matters only in relation to that lost constraint. If an AI-native firm offers a genuinely different product to a different group of buyers, it may not replace the rivalry between Seismic and Highspot. If it reaches the same customers and forces both legacy providers to improve or moderate their terms, it may be a meaningful constraint even before it becomes a large incumbent. The legal and economic analysis must distinguish those possibilities rather than treating the market as a single technology race.
Why the relevant market comes first
A relevant market is the arena in which the alleged competitive effect has meaning. It normally has a product dimension and a geographic dimension. The product question asks which offerings customers regard as reasonably interchangeable for the purposes at issue. The geographic question asks where customers can realistically turn to competing suppliers. Neither dimension is merely a label.
The Tapestry–Capri opinion makes the sequencing explicit. The accessible court record says that a plaintiff must propose a proper relevant market and show that the merger’s effects in that market are likely to be anticompetitive. It describes market definition as a necessary predicate because the significance of a merger depends on the market affected. The court also treated the 2023 Guidelines as a persuasive tool rather than binding law. That distinction matters: agency guidance can organize analysis, but a court still applies the governing statute and evaluates the facts before it.
Market definition also protects against two opposite errors. A market can be drawn too narrowly, making ordinary alternatives disappear from view and exaggerating concentration. It can also be drawn too broadly, grouping together products that customers do not actually treat as substitutes and understating the loss of rivalry. The point is not to select the most favorable boundary for either side. It is to locate the competitive constraint that the merger allegedly removes.
This is why a technology category is not automatically a market. “AI software,” “enterprise software,” or “sales enablement” may each contain products that differ in customer, function, route to market, and switching cost. A broad category can be useful for orientation, but a merger concern requires a more disciplined question: if one supplier worsened its terms or stopped improving, which alternatives could the affected customers actually use? The answer may include products that do not share the same technical architecture, and exclude products that use similar technology but do not compete for the same decision.
The accessible Tapestry record illustrates how ordinary business evidence can bear on that boundary. Docket-level summaries refer to customer switching under the hypothetical-monopolist test, Capri’s own descriptions of the MICHAEL Michael Kors line as accessible luxury, internal distinctions between affordable-luxury and luxury brands, and a dispute over whether excluding wholesale data distorted price-band calculations. The lesson is not that handbag evidence transfers directly to enterprise software. It is that market categories must be tested against customer behavior, company practice, and the data used to measure the market.
For Seismic and Highspot, the corresponding questions would be concrete. Which businesses buy sales enablement software? What features or workflows make one platform a substitute for another? Can customers move between platforms without prohibitive costs or lost functionality? Do AI-native providers serve the same buyers, or do they offer an adjacent tool that still leaves the central rivalry intact? The public materials do not answer all of these questions. That limit is part of the analysis: a reader can identify the questions without pretending that the public release resolves them.
From a prima facie concern to a rebuttal
Once a plaintiff defines the relevant market and presents evidence that the merger is likely to harm competition there, the analysis reaches what courts and agencies often call a prima facie case—a showing sufficient to support an initial inference of likely harm. In the accessible Tapestry opinion, the court describes a burden-shifting framework. The plaintiff must establish a proper market and likely anticompetitive effects. If that showing succeeds, the defendant can present evidence that the prediction is inaccurate or that the evidence supporting it is sufficiently weakened. The ultimate burden of persuasion remains with the plaintiff.
This sequence helps clarify the role of disruption evidence. A company defending a merger is not necessarily required to prove that every future competitor will be absent. It must show why the evidence about the transaction’s likely effect is incomplete or misleading, or why other competitive forces will prevent the predicted harm. A future entrant can be part of that response, but its existence as a possibility is not enough.
For Seismic and Highspot, that distinction prevents a common shortcut. The defense need not predict the entire future of software, and the Division need not prove that no AI-native firm could ever enter. The relevant question is narrower: given the competitive relationship the transaction would remove, is the evidence strong enough to conclude that other forces will prevent a substantial reduction in competition? A general forecast about faster innovation may be consistent with that conclusion, but it does not establish it. The forecast has to be connected to customers, alternatives, and the timing of competitive response.
The 2023 Guidelines describe entry or repositioning as rebuttal evidence when it is timely, likely, and sufficient in its magnitude, character, and scope to deter or counteract the competitive effects of concern. “Timely” means fast enough to replace lost competition before the merger can produce the threatened effect, and durable enough that a short-lived experiment does not count as long-term protection. “Likely” asks why the entrant would actually enter under the conditions expected after the merger, especially if it had not planned to enter under pre-merger conditions. “Sufficient” asks whether the response can match the scale, strength, and durability of a merging party’s competitive role.
The order of those questions matters. A possible entrant that will arrive only after a lengthy product build, customer-by-customer implementation cycle, and uncertain funding path may be too slow even if its eventual product would be strong. An entrant that could launch quickly but has no clear incentive to invest under expected post-merger conditions may be too speculative. And a fast, motivated entrant may still be insufficient if it can serve only a narrow slice of the customers or functions that gave the merging firms their competitive significance. The standard is demanding because the rebuttal must protect competition, not merely show that the industry will remain active.
These are separate tests because a response can pass one and fail another. An AI-native provider might be technically ready but slow to win enterprise customers. It might win customers quickly but lack the durability to sustain investment. It might be durable but too narrow to constrain the features or buyers that made Seismic and Highspot close competitors. Or it might be strong in one dimension of competition while failing to replace the full role of the merged firms. Calling all of these possibilities “disruption” hides the decision that the evidence must support.
The Guidelines also distinguish a potential-entry concern from entry as rebuttal. A merger can be harmful because it eliminates a firm that might have entered or a firm whose perceived ability to enter already pressures incumbents. That does not mean a defendant can rebut a current-competitor concern by pointing to the mere possibility of another entrant. The likelihood needed to establish that a potential entrant matters at the incipient stage is generally lower than the likelihood needed to show that entry will neutralize a demonstrated risk of substantial lessening.
Applied to Seismic and Highspot, this model yields a restrained conclusion. The public record makes it reasonable to study AI-native entry because the Division says those firms were winning customers and increasing pressure. It does not allow a reader to skip the preceding steps or to claim that AI-native growth automatically neutralized the merger. The relevant market, the lost constraint, the timing of customer adoption, and the durability and scope of the pressure still determine what the evidence means.
The next question is therefore more specific: how does an AI-native firm move from a technical possibility or potential entrant into a competitive force capable of replacing what a merger removes?