Brazil Treats Minority AI Stakes as Major Competition Signals

The defining antitrust question of the artificial intelligence era may be less about how aggressively regulators intervene than how long they can afford to wait.

And Marcela Mattiuzzo, partner at VMCA Advogados, told Competition Policy International (CPI), a PYMNTS company, that Brazil is emerging as a test case. The state’s competition authority, CADE, has moved quickly in digital-market investigations, scrutinized unconventional AI transactions and sought more information about investments connecting technology incumbents with emerging AI companies.

“Some of these authorities have the perception that they missed the target on digital markets,” Mattiuzzo explained during a discussion for the CPI TechREG talks series. Regulators, she added, increasingly believe “they should have changed their minds faster, and they should have done something sooner.”

AI is arriving just as authorities are trying not to repeat that experience.

The Cost of Waiting Is Changing the Antitrust Risk Equation

Antitrust enforcement has traditionally carried an asymmetric risk: intervening incorrectly can damage competition, while dynamic markets may sometimes correct competitive problems themselves. Fast-moving AI markets complicate that calculus because the market can evolve faster than an investigation.

CADE’s use of interim measures reflects that tension. Without an ex ante digital competition regime, Brazilian regulators are relying on existing tools when they believe waiting could allow competitive conditions to become difficult to reverse.

Mattiuzzo noted that CADE appears to believe the risks of interim intervention “are smaller when compared to the risk of not doing anything.”

The significance is not around more aggressive enforcement but earlier scrutiny instead. Regulators are now asking whether transactions and commercial relationships that appear relatively small today could influence tomorrow’s market structure.

“There’s an asymmetry that is huge between companies and the authorities,” Mattiuzzo explained.

Companies have product strategies and business plans that offer some view of where they expect markets to develop. Regulators must assess those possible futures from outside alongside any technology, talent, infrastructure relationships or distribution qualities that regulators can foresee as becoming strategically important.

“Making any futurology here is particularly hard, because we have very little clarity,” Mattiuzzo said, adding that the reality could turn five-year projections, integration strategies, financing relationships and infrastructure dependencies into important evidence in competition reviews as antitrust analysis moves upstream, potentially before revenue or even the ultimate business model has stabilized.

The challenge is therefore not simply moving faster. It is building an enforcement model capable of learning quickly while retaining predictable limits.

Brazil is “well positioned to be in this conversation,” Mattiuzzo said, although she cautioned against assuming that moving first is necessarily an advantage. “There’s a very high chance that you end up doing something that is not necessarily workable, and then you have to change.”

Minority Stakes No Longer Look Minor in the AI Era

Among the key tensions defining the next phase of AI antitrust is the fact that regulators spent much of the previous technology cycle concerned about acting before they understood the market. In AI, they are confronting the opposite possibility: by the time the market is fully understood, some of its most important competitive choices may already have been made.

“Minority investments might become more relevant in this context than they have usually been, at least in Brazilian competition practice,” Mattiuzzo said.

Large technology companies, she added, can establish significant relationships with emerging AI developers through minority investments, cloud agreements, licensing arrangements and talent deals without completing traditional acquisitions.

That does not necessarily mean more transactions will be blocked. It could mean more notification, information requests and scrutiny as authorities attempt to map relationships across capital, compute, models and distribution. After all, in AI, competitive significance can emerge before financial significance.

Brazil generally gives CADE a 12-month window to call in certain transactions that were not subject to mandatory notification. Mattiuzzo argued that boundary should remain meaningful.

“We should try and stick to that,” she said. “Otherwise, we create a whole bunch of problems.”

The AI regulatory nightmare, at least in Brazil, is shifting from the false positive, or stopping behavior that might have proved harmless; toward the false negative of allowing an ecosystem to consolidate before regulators understand where competitive power is accumulating.

Watch the full TechReg Talks episode with VMCA Partner Marcela Mattiuzzo to hear more about:

• Why AI is changing the cost of waiting in antitrust. Mattiuzzo says competition authorities are reacting to a belief that they moved too slowly in earlier digital markets, making them more willing to intervene before AI markets tip or competitive positions become difficult to unwind.
• Why business plans and minority investments are becoming antitrust evidence. As regulators try to assess what young AI companies could become, five-year projections, cloud relationships, licensing deals and minority stakes may matter more than current revenue or formal control.
• Why faster enforcement still needs a clear legal boundary. Mattiuzzo argues regulators need earlier visibility into emerging AI markets, but warns that open-ended review creates its own risks, making predictable limits such as Brazil’s 12-month call-in window critical to preserving legal certainty.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.