Buying the Future to Kill It

Big Tech acquisitions can be efficient deals that damage the system that made them valuable

The paper’s central case is the Hollow Win: C=0, A=1, B=1. The acquirer gains, the startup’s shareholders gain, and the innovation ecosystem loses future competitors, funding, and entrepreneurial labor.

• Facebook bought Instagram on April 9, 2012 for $1 billion when Instagram had 13 employees and no revenue.
• The FTC cleared the deal under Hart-Scott-Rodino review in 30 days because the frame was photo-sharing market concentration.
• By 2023, Instagram had more than 2 billion monthly active users and generated an estimated $43 billion in advertising revenue.

The regulator sees acquirer and target, but not system C

Standard antitrust review models the transaction around acquirer A, target B, and a regulator using consumer-welfare proxies. The paper argues that this payoff space cannot record ecosystem damage.

• The bilateral game G= A,B yields four outcomes; the transformed game G1= A,B,C yields eight.
• The missing outcome is C=1, A=1, B=1: acquirer, target, and ecosystem all gain.
• The canonical Big Tech outcome is C=0, A=1, B=1 because neither transaction party pays for system-welfare loss.

Instagram became a co-specialized Meta asset instead of an independent competitor

The paper does not claim Instagram was shut down. It claims the acquisition converted a future competitive threat into a platform asset and changed entry expectations around social media.

• FTC v. Facebook documents cited in the paper report Zuckerberg viewed Instagram as a company that could grow into a Facebook competitor.
• VC investment in social media startups within two product spaces of Meta platforms fell an estimated 46% in the three years after the acquisition.
• Social media startups founded per year fell from 127 in 2011 to 43 in 2016, according to the paper’s CB Insights citation.

Foreclosure works through ambiguity, switching costs, and the kill zone

The paper’s mechanism is structural foreclosure of future entry, not project termination. The acquired product can grow while the surrounding startup market thins out.

• Causal ambiguity: integration lets the platform blur whether the acquired technology improves the ecosystem or blocks future rivals.
• Co-specialization: Instagram users’ social graph, content history, and connected services become embedded in Meta infrastructure.
• Kill zone: Kamepalli, Rajan, and Zingales report VC investment drops 46% and total deals drop 20% after major platform acquisitions.

A Pareto-optimal acquisition can still lower system welfare

The Missing System Theorem gives the paper its formal spine: bilateral optimality does not imply system welfare. The welfare dimension W is orthogonal to the payoffs of A and B.

• A strategy profile s* can be Pareto-optimal for A and B while W(s*) is lower than W(s') for a Pareto-dominated alternative.
• In acquisitions, the visible numbers are price, premium, integration plan, and expected acquirer gain.
• The invisible cost is the future competition that no longer enters the market.

The paper estimates β_W=7.81 for Big Tech acquisition revenue

Using the System Asset Pricing Model, the paper defines β_W=-dW/dΠ. For Big Tech acquisitions, it estimates that each dollar of acquisition-related revenue destroys $7.81 of system welfare.

• Big Tech acquisition-related revenue is estimated at $128 billion.
• Annual welfare loss is estimated at $999.6 billion, with a 90% confidence interval for β_W of [6.00, 9.60].
• The estimate combines foreclosed innovation value, reduced consumer surplus, and ecosystem degradation.

The $999.6 billion estimate comes from three named loss channels

The paper breaks the annual welfare cost into components tied to VC suppression, lost product-market competition, and weaker ecosystem capacity.

• Foreclosed innovation value: 420 billion, based on a 46% VC reduction applied to platform-adjacent investment from 912 billion total VC in 2023.
• Reduced consumer surplus: $310 billion, using a conservative 3% Big Tech killer-acquisition rate adapted from Cunningham, Ederer, and Ma’s 6% pharmaceutical estimate.
• Ecosystem degradation: $269.6 billion from talent exit, weaker spillovers, and R&D redirected toward incumbent-friendly paths.

Each acquisition narrows the next set of competitive futures

The Kill Zone Ratchet Theorem formalizes why the cost compounds. Once a potential competitor is acquired, the market does not return to its prior set of feasible competitive states.

• Acquisition is an absorbing state because the acquired firm cannot become an independent competitor under the acquirer’s control rights.
• Each acquisition expands the product spaces where entry is deterred by the platform’s capabilities and market presence.
• When acquisitions are profitable, the kill zone expands, and regulators evaluate bilaterally, system welfare decreases as acquisitions accumulate.

WhatsApp, Waze, and GitHub repeat the Hollow Win structure

The paper uses four cases to separate deal-level gains from system-level loss. Each acquired product continued, but the surrounding competitive field weakened.

• WhatsApp: Facebook paid $19 billion in 2014 when WhatsApp had 450 million users; by 2020 it had 2 billion users, while data sharing with Facebook began in 2016.
• Waze: Google paid $1.1 billion in 2013 for a community navigation app with 50 million users; no new navigation startup later reached competitive scale against Google Maps.
• GitHub: Microsoft paid $7.5 billion in 2018 when GitHub had 28 million developers; by 2023 GitHub had 100 million developers, with governance shifting toward Microsoft’s developer ecosystem.

The paper rejects blanket bans and targets foreclosure value

The proposed game change is conditional licensing, not a ban on acquisitions. The rule lets platforms integrate technology while removing the private payoff from locking the ecosystem out.

• The paper cites a 2026 CEPR/LSE working paper finding that 70% of platform acquisitions are cross-industry and that blanket bans could reduce welfare by 0.21%.
• It cites Letina, Schmutzler, and Seibel to warn that blocking genuine technology acquisitions can reduce innovation incentives.
• The proposed fix conditions acquisitions so technology value remains available while foreclosure value is stripped out.

Designated platforms must keep acquired startup assets available to competitors

The paper’s rule applies when a designated platform acquires a startup. It changes the acquisition terms so the ecosystem can still use the acquired technology, data access, interoperability, and talent market links.

• Technology licensing: license patents, copyrights, and trade secrets needed to replicate core functionality on FRAND terms for five years.
• Data portability: let users of the acquired product move their data to competing platforms at no cost.
• Interoperability and non-exclusivity: maintain interoperability for three years and bar exclusivity for employees, contractors, and partners for two years.

The Digital Markets Act supplies the administrability model

The paper points to the EU Digital Markets Act, effective March 2024, as evidence that gatekeeper obligations can be specified and enforced without predicting every future competitive path.

• The DMA designates Alphabet, Amazon, Apple, ByteDance, Meta, and Microsoft as gatekeeper platforms.
• Its obligations include interoperability, real-time data portability, and non-discrimination for core platform services.
• The paper’s extension is acquisition-specific: when a gatekeeper buys a startup, the acquired technology must remain available to the ecosystem.