Amazon, whose stock price crossed $1,000 per share last week, illustrates the shortcomings of our current antitrust regime.

Last Tuesday, Amazon’s stock price crossed $1,000 per share, reaching an all-time high for the company. Wall Street analysts had predicted the event after Amazon reported 23 percent growth in sales and $35.7 billion in revenue as part of its first-quarter earnings, showcasing both its established dominance and continued expansion. Regardless of how much or little significance one attributes to the $1,000-level itself, the occasion marked another milestone for a company whose expanding reach and growth across our economy has been astonishing. As the New York Times observed, “Cracking the arbitrary threshold of a four-digit stock price is a reminder of just how heavy Amazon has become.”

Indeed, there is growing recognition that Amazon now plays a leading role across the Internet economy—and beyond. In 2016, it captured 43 percent of all the revenue generated online in the U.S. and 53 percent of online sales growth, with its share only predicted to increase. Retail stores are shuttering at an accelerated pace; what had been a steady decline now looks like decimation. And in addition to being a retailer, Amazon is now a delivery and logistics service, a credit lender, a hardware manufacturer, a producer of television and films, a book publisher, and a major host of cloud computing services. Not only is Amazon the titan of e-commerce but its lead may be “uncatchable.”

In “Amazon’s Antitrust Paradox,” recently published at the Yale Law Journal, I argue that elements of Amazon’s structure and conduct raise anticompetitive concerns. Critically, however, the current framework in antitrust fails to grasp the architecture of power in the platform economy. By viewing competition primarily through the prism of short-term price effects, the “consumer welfare” approach to antitrust ignores real harms to competition. Amazon illustrates the shortcomings of this regime.

Key to understanding how Amazon has sidestepped antitrust law is tracing its path of growth. Amazon established dominance as an online platform due to two elements of its business strategy: a willingness to sustain heavy losses and invest aggressively at the expense of profits, and integration across multiple business lines. For the majority of its twenty years in business, Amazon generated negative net income.((Due to the success of Amazon Web Services, Amazon has now begun to report consistent profits. But the losses Amazon formerly undertook are still significant for understanding how it established a dominant position.)) Even when it did enter the black its margins were razor-thin, despite staggering growth. In the article, I review how Amazon established its Prime loyalty program, entered into the e-book market, and undertook its acquisition of Quidsi. In each case Amazon achieved its position through deeply cutting prices below-cost or investing heavily in expanding operations—both at the expense of profits.

Although laws prohibiting predatory pricing were an antitrust mainstay through the 1960s, in the late 1980s and early 1990s the Supreme Court introduced and adopted a test that dramatically raised the burden of proof. Driving the court’s analysis was the idea that “predatory pricing schemes are rarely tried, and even more rarely successful.”

But platform markets—characterized by network effects and the self-reinforcing advantages of data insights—are won through establishing an early lead, by chasing market share and driving out rivals. Due to this dynamic, seeking to maximize market share at the expense of one’s competitors makes predatory pricing highly rational; indeed, it would be irrational for a firm not to endure losses to capture the market. The trajectory of Amazon’s stock price—which has long been untethered from reported profits—shows that investors recognize this fact: in platform markets, bleeding money to establish growth may be a guarantor of long-term dominance. The market reflects a truth that our current laws fail to detect.

The other critical facet of Amazon’s strategy is integration, which permits Amazon to leverage advantages across business lines. In the article I also assess how Amazon Marketplace and Fulfillment-by-Amazon have emerged as infrastructure for the internet retail economy, and how Amazon’s decision to retail goods directly while also serving this infrastructure role creates significant conflicts of interest.

For example, Fulfillment-By-Amazon demonstrates how the company harnessed its dominance as an online retailer to integrate into delivery, in part through enjoying significant bargaining power with FedEx and UPS. It also highlights conflicts of interest, where Amazon is positioned to use its logistics infrastructure to deliver its own retail goods faster than those of independent sellers. Its Marketplace for third-party sellers similarly underscores this risk, as Amazon uses sales data from independent merchants to make its own retailing decisions and to undercut other sellers on price, while giving its own items featured placement in search results. It has used similar tactics when establishing its private brand, deploying insights gleaned from third-party sellers to directly produce goods other merchants had originally sold.

By harnessing data collected through its platform in this way, Amazon increases its own sales while outsourcing the risk undertaken by independent sellers. Reports suggest Amazon also leverages insights from Amazon Web Services to benefit other lines of business: it has made investment decisions based on which start-ups are most rapidly expanding their usage of AWS. That third-party businesses depend on Amazon’s infrastructure services means the company can use these tactics without any realistic fear of losing significant business.

While antitrust law traditionally viewed integration by dominant firms as ripe for anticompetitive abuse, the Chicago School treats integration as largely benign, on grounds that monopoly leveraging—if it occurs—is likely to generate efficiencies. Any temptation to abuse power attainted through integration, meanwhile, is assumed away by the constant threat of potential entry. Experience shows that this assumption is misguided generally and especially so in platform markets, where network effects and the self-reinforcing advantages of data create significant barriers to entry.

Amazon’s scope and depth of power across the retail economy is already vast. As it continues to expand, deepening its existing control over infrastructure and reaching into new lines of business, we will need to decide how law should grapple with anticompetitive facets of its dominance—as well as with the power of other dominant platforms, including Google, Apple, and Facebook.

A key question is whether to govern online platform markets through competition, or to accept that they are inherently monopolistic and regulate aspects of their power instead. Choosing competition will require that we reform antitrust law to prevent this dominance from emerging—possibly by introducing a presumption of predation for dominant platforms, adopting a prophylactic ban on integration that creates significant conflict of interest, and recognizing the anticompetitive prospect of mergers that involve acquisitions of valuable data and opportunities to cross-leverage it.

Accepting these platforms as natural monopolies, meanwhile, could involve adopting a nondiscrimination policy to prevent platforms from privileging their own goods and from discriminating among platform users and applying either the essential facilities doctrine or common carrier obligations to ensure open and fair access to other businesses.

Revising our legal framework to capture the reality of how dominant firms acquire and exercise power in platform markets is critical. Thankfully, there is growing recognition among public commentators, politicians, and policymakers that the present approach to competition policy has failed to preserve competitive markets, and that the current paradigm should be reformed or even abandoned. How this reform movement addresses the power of dominant platforms will prove a key test.