The Invisible Checkpoint

It's Monday, April twenty seventh, and welcome to HigherIQ. Last week, SpaceX announced it had secured the right to acquire Cursor, the AI coding tool, for sixty billion dollars. A few days earlier, Microsoft had looked at buying Cursor and decided not to bid. The deal hasn't closed yet. It might never close. Because before any acquisition of that size is finalized, it has to pass through an invisible checkpoint that most people outside of corporate law don't know exists. The United States government has the power to kill a deal between two willing parties, two companies that want to merge, have agreed on a price, and are ready to sign, if it decides that the deal would reduce competition. The law that gives them that power was written in 1890 to break up Standard Oil. It was used in 2020 to kill Visa's acquisition of Plaid. It was used in 2023 to kill Adobe's acquisition of Figma. And the theory behind those recent cases represents a fundamental shift in how the government thinks about competition: they're no longer just blocking deals that hurt consumers today. They're blocking deals that might eliminate competitors that don't fully exist yet. The Sherman Antitrust Act was signed into law on July 2, 1890, by President Benjamin Harrison. It was written in response to the trusts, massive corporate combinations that had come to dominate entire industries in the decades after the Civil War. The most infamous was John D. Rockefeller's Standard Oil Trust, which by 1900 controlled ninety percent of the oil refining business in the United States. The Act is remarkably short. Section 1 declares illegal "every contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade." Section 2 makes it a crime to "monopolize, or attempt to monopolize" any part of interstate commerce. The law was largely ignored for its first decade. Then Theodore Roosevelt used it to break up the Northern Securities railroad monopoly in 1904, and in 1911, the Supreme Court ordered the breakup of Standard Oil into thirty four separate companies. Some of those companies became Exxon, Mobil, Chevron, and Amoco. The same law was later used against AT&T in 1982, forcing the telecommunications giant to split into seven regional "Baby Bells." Microsoft was sued under the Sherman Act in 1998 for bundling its web browser with Windows, though the company avoided breakup through a settlement. In 1914, Congress added the Clayton Act and the Federal Trade Commission Act, which specifically addressed mergers and acquisitions. The Hart-Scott-Rodino Act of 1976 added a procedural requirement: any merger above a certain size (currently about one hundred and eleven million dollars) must be reported to the DOJ and FTC before it can close. The agencies then have thirty days to review the deal, and can extend that review by issuing a "Second Request" for more information, a process that typically takes six to twelve months and costs the companies millions in legal fees. This is the checkpoint. Every major acquisition in America passes through it, and the government can block the deal if it determines that the merger would substantially lessen competition. For most of the twentieth century, antitrust enforcement focused on deals between existing competitors. If the number one and number two companies in an industry wanted to merge, the government would likely block it because the combined company would have too much market power. The math was relatively straightforward: measure market share, assess the competitive landscape, predict the effect on prices. The deals that have defined the current era of antitrust are different. Visa's attempted acquisition of Plaid in 2020 and Adobe's attempted acquisition of Figma in 2022 both involved a dominant company buying a smaller company that wasn't yet a direct competitor but could become one. Visa announced it would buy Plaid for 5.3 billion dollars in January 2020. Plaid was a fintech company that connected consumer bank accounts to financial apps like Venmo and Robinhood. It wasn't competing with Visa in payment processing. But the DOJ argued that Plaid was building technology that would eventually allow it to challenge Visa's dominance in online debit transactions. Visa's own CEO had acknowledged in internal discussions that Plaid could create a competitive threat. The DOJ's theory was what antitrust lawyers call "nascent competition": Visa wasn't buying a competitor. It was buying a company specifically to prevent it from becoming one. The DOJ called it a "killer acquisition." Visa pulled the deal in January 2021 rather than fight the case in court. Adobe's twenty billion dollar bid for Figma followed a similar pattern. Adobe dominated professional design software with Photoshop, Illustrator, and its Creative Cloud suite. Figma was a browser-based design tool that was gaining market share rapidly, particularly for collaborative interface design. The DOJ and the European Commission both investigated the deal and concluded that it would eliminate a company that was on track to become Adobe's most significant competitor. Adobe abandoned the acquisition in December 2023 and paid Figma a one billion dollar breakup fee for its trouble. The pattern in both cases is the same: a dominant company identifies a smaller company that threatens its future position, offers a price so high the smaller company can't refuse, and tries to absorb the threat before it fully materializes. The government's job, under the nascent competition theory, is to spot this strategy and block it, even though the two companies aren't competitors yet. The argument is that if you only block mergers between existing competitors, you'll never catch the deals that prevent competition from emerging in the first place. The counterargument is real and worth taking seriously. If the government blocks too many acquisitions, it reduces the number of exit opportunities for startups and their investors. Venture capital depends on the expectation that successful startups can eventually be acquired by larger companies. If founders and investors believe the government will block any deal involving a large buyer, they may invest less in startups, which could reduce innovation rather than promote it. Microsoft's sixty nine billion dollar acquisition of Activision Blizzard in 2023 went through after the company made concessions on cloud gaming rights, suggesting that not every large deal gets killed. The line between legitimate acquisition and killer acquisition isn't always clear. This brings us back to SpaceX and Cursor. SpaceX, which merged with Elon Musk's AI company xAI in February, has secured an option to buy Cursor for sixty billion dollars later this year. The structure is notable. It's not a completed acquisition. It's a collaboration agreement with a purchase option attached. SpaceX and Cursor will work together to develop AI coding models using xAI's Colossus supercomputer, and at the end of the collaboration, SpaceX can either pay ten billion for the joint work or exercise the option and acquire the company outright. This structure may be designed partly to manage regulatory timing. SpaceX is planning what could be the largest IPO in history, targeting a valuation between 1.75 and 1.8 trillion dollars in June. Announcing a clean acquisition of a company in a hot market while simultaneously preparing an IPO creates regulatory complexity. An option agreement defers the merger review until after the IPO closes. But the antitrust question will eventually arrive. Cursor is the dominant AI coding tool. SpaceX, through xAI, is building AI models that compete with the very models Cursor currently sells access to, including Claude and GPT. If SpaceX acquires Cursor, it could eventually replace those third-party models with its own, creating a vertically integrated AI coding stack. Microsoft, which owns GitHub Copilot, the main competitor to Cursor, looked at buying Cursor and chose not to. The DOJ might ask whether allowing SpaceX to acquire Cursor would reduce competition in the AI coding market, particularly given Musk's expanding control over AI infrastructure, social media, and now potentially the most popular developer tool in the world. Whether regulators challenge this deal will depend on how they define the relevant market, how much competition exists among AI coding tools, and whether they apply the same nascent competition theory that killed the Visa and Adobe deals. The law is the same one that broke up Standard Oil in 1911. The question is whether a statute written to regulate railroad monopolies can effectively govern a market where the most valuable companies are less than five years old. So if this comes up in conversation, here's how to think about it. Every major acquisition in the United States has to be reported to the government before it can close, and the DOJ or FTC can block it if they determine it would substantially reduce competition. The Sherman Antitrust Act of 1890, the same law used to break up Standard Oil, is still the foundation. What's changed is the theory: the government is now blocking deals not just between existing competitors but between a dominant company and a smaller company that might become a competitor in the future. Visa tried to buy Plaid to prevent it from threatening its debit business. Adobe tried to buy Figma to prevent it from threatening its design software dominance. Both deals died. The SpaceX and Cursor deal hasn't reached that checkpoint yet, but it will. The question the government will ask is the same one it's been asking since 1890: does this deal make the market more competitive or less? The law is old. The companies are new. The question hasn't changed. Stay informed, stay curious, and we'll see you tomorrow.