Why the Platform Giant Losing Money Forever Is Not Doing Anything Wrong

Based on the research ofLiang Chen, "Ecosystem Competition and Cross-Market Subsidization: A Dynamic Theory of Platform Pricing," arXiv preprint, 2026

The Puzzle That Predation Doctrine Cannot Explain

Antitrust law has a clean test for pricing that looks too good to be true. Under the U.S. Supreme Court's 1993 ruling in *Brooke Group Ltd. v. Brown & Williamson Tobacco Corp.*, a plaintiff alleging predatory pricing must show two things: that prices sat below an appropriate measure of cost, and that the predator had a "dangerous probability" of recouping those losses later through higher prices once rivals were driven out. The logic is sound as far as it goes. If a firm cannot plausibly recoup, then bleeding money is just a subsidy to consumers, not a competitive weapon, and the antitrust system has no business punishing generosity.

The trouble, according to a new theoretical paper by Liang Chen, is that a whole class of real platform pricing never fits either box. Chen's dynamic game, posted to arXiv in January 2026, starts from an empirical puzzle: platform giants operating in highly concentrated markets, some with over 60 percent share in their core segment, have sustained compressed margins for years, not months. Standard industrial-organization theory predicts one of two endings for that situation. Either the weaker rival eventually exits and the survivor raises prices toward monopoly levels, or the low pricing is proven predatory and stopped. Neither ending arrives. The margins stay thin indefinitely, the dominant firm stays dominant, and no recoupment phase ever shows up on the income statement.

Chen's answer is to stop modeling the firm as an optimizer of one market and start modeling it as an optimizer of an ecosystem, a portfolio of markets a single user touches across a single day. This is a theoretical, model-based paper, not a field study; it reports no new experimental or transaction-level data of its own. What it offers instead is a mechanism, expressed as a dynamic game with a proven equilibrium, for why the puzzle resolves once the unit of analysis shifts from the market to the ecosystem.

The Mechanism: Spillovers Change What "Profitable" Means

The formal move in the paper is to define what Chen calls ecosystem complementarity: the spillover value a user active in one market generates for the firm in an adjacent market it also operates. A ride-hailing user who also orders food through the same app, pays through the same wallet, and watches ads served against the same profile is worth more than the fare or delivery fee alone. Standard single-market pricing ignores that spillover. Chen's model prices it in.

The dynamic game shows that once this complementarity crosses a threshold, the firm's individually rational move is to keep subsidizing the loss-making market forever, not just until a rival capitulates. Losing money on rides is worth it indefinitely if the rides keep generating payments users, payments users keep generating retail or delivery users, and the whole bundle keeps generating data and advertising value that no single market's price reflects. The paper characterizes this outcome as the unique stable equilibrium under strong-enough complementarity, and shows it is dynamically stable, firms that deviate toward cost-covering prices lose ecosystem share and have an incentive to revert, so the below-cost price does not erode over time the way a temporary price war would.

That stability is the paper's most consequential claim. A temporary price war ends because it is expensive and someone runs out of capital. A stable equilibrium does not end on its own, because ending it is not in anyone's individual interest even though the collective outcome, capital pouring into subsidies year after year instead of new products, is what the paper calls potentially inefficient in aggregate.

What This Looks Like Off the Page

Chen's model is theoretical, but the shape of the equilibrium it describes, years of subsidized competition between ecosystem players, funded by capital rather than category profit, with no clean recoupment moment, is recognizable in several real, well-documented cases, none of which the paper itself analyzes.

The clearest example is the 2015-2016 ride-hailing war between Didi Chuxing and Uber China. Didi and its earlier rival Kuaidi Dache had already burned through roughly $700 million combined in an even earlier subsidy war before merging in 2015. The fight then escalated with Uber: Uber lost more than $1 billion in China in 2015, and Didi's overall losses that year reportedly reached $1.8 billion, according to the Chinese business publication *Caixin*. Both companies raised outside capital specifically to fund the subsidies, not ride-hailing profits, which did not exist. The war ended not with a price recovery but with a merger: Didi absorbed Uber China in August 2016 in a deal valuing the combined company near $36 billion, with Uber taking roughly a fifth of the merged entity's equity. There was no recoupment phase in the Brooke Group sense, Uber never raised prices to extract back its loss, because it exited the market and took a stake in its former rival instead. That is a capital-funded subsidy contest resolving through consolidation, not through one side proving it could raise prices later.

A newer instance is playing out now. In 2025, JD.com's push into food delivery ignited a subsidy war with Meituan and Alibaba's Ele.me and Taobao Instant Commerce, with daily order volume on some platforms reportedly surpassing 80 million during the fighting. China's State Administration for Market Regulation summoned the three companies in July 2025, and by August all three had issued public pledges to curb the practice, Meituan explicitly committing not to price "significantly below cost," and JD.com pledging not to use transaction volume as a proxy for market dominance. Notice what the regulator actually targeted: not a predation lawsuit over any single price, but a negotiated, multi-party commitment about subsidy conduct going forward, closer to Chen's prescription of intervening on cross-market subsidization than to a Brooke Group-style case built around one firm's price and a hunt for recoupment.

The regulator did not sue anyone for predatory pricing. It summoned three ecosystem players and asked them, together, to stop subsidizing so hard. That is targeting the capital flow, not the price.

A third pattern, less dramatic but longer-running, is visible in how Amazon's segments have reported profit for a decade. Amazon's International retail segment posted an operating loss every year from 2015 through 2022, roughly $22.5 billion cumulative, while Amazon Web Services generated the majority of the company's total operating income across most of that same period, according to Amazon's own segment disclosures compiled in financial reporting. This is not evidence of predation in any market AWS competes in; cloud computing and international retail are not rivals. It illustrates the more general mechanism the paper formalizes: profit generated in one part of an ecosystem can fund sustained losses in another part for years, as a matter of ordinary capital allocation, without any single market's price ever needing to make sense in isolation.

The Regulatory Redirection the Paper Argues For

Chen's policy recommendation follows directly from the mechanism: since the equilibrium is stable and individually rational, chasing it with predatory-pricing suits aimed at a single market's prices will keep failing to find a recoupment phase, because there structurally isn't one to find. The paper instead proposes three levers aimed at the ecosystem level rather than the price level: data-portability mandates that reduce the switching cost locking users into the spillover loop, restrictions on intra-ecosystem capital transfers that fund one market's losses from another market's profits, and transparency requirements around customer-acquisition cost so outside observers can actually see the subsidy rather than infer it from a suspiciously low price.

The European Union's Digital Markets Act already has a working version of the first lever. Article 6(9) requires designated gatekeeper platforms to give end users, and any third party the user authorizes, continuous and real-time portability of the data the user provided or generated on the platform, delivered free of charge. The mechanism is not about pricing at all, but it attacks the same spillover the model treats as the source of the subsidy's payoff: if a user's data and history move freely to a rival, the value of keeping that user captive inside one ecosystem drops, and with it the return on subsidizing them there in the first place. That is Chen's first recommended lever, already implemented, aimed precisely at the variable the model says drives the equilibrium.

The second and third levers, capital-transfer restrictions and acquisition-cost transparency, have no comparable real-world precedent yet. That gap is itself informative. Regulators have built tools to weaken the spillover (portability) but not yet tools to see or restrict the capital flow the spillover funds. Chen's paper is, in effect, an argument that the second half of the toolkit is missing and that its absence is why persistent below-cost pricing by ecosystem giants keeps looking like an unsolved puzzle to observers using a single-market lens.

What to Watch Instead of the Price

The practical takeaway for anyone competing against, regulating, or simply trying to model an ecosystem platform is to stop asking whether today's price makes sense in its own market. Under Chen's model, it may never need to, and it may never be intended to recoup. The question worth asking instead is where the subsidy is funded from, how strong the spillover is that makes funding it worthwhile, and whether anything is currently making that spillover weaker, through portability, interoperability, or disclosure. A stable equilibrium does not announce itself with a price spike after the rival leaves. It just continues, indefinitely, looking exactly like healthy competition from inside any one of its markets.