The walled garden
Five firms hold the substrate of the consumer internet (cloud, mobile OS, social-graph, search, e-commerce) and take rent on every layer above. The 25-year US antitrust gap between Microsoft 2001 and the 2024 Google ruling produced the concentration; the EU has been the de-facto regulator over the same window. Privacy law in the United States is a patchwork of state statutes; the federal framework is 1996 (HIPAA) and 1998 (COPPA). The 2022 tech competition bill that would have changed any of this died after $120M in industry lobbying. Sources are FTC, DOJ, EU Competition Commission, and the firms' own 10-K filings.
Concentration
4 itemsSeven firms (Apple, Microsoft, Alphabet, Amazon, Nvidia, Meta, Tesla) carry ~$15T in combined market cap, more than the UK, Japanese, French, and Indian stock markets put together. The S&P 500 is now a Magnificent-Seven concentration index masquerading as a broad index. Google holds ~90% of US search; AWS+Azure+GCP control ~65% of global cloud; Meta-owned platforms hold ~45% of US social-media time. These are not adjacent monopolies; they are the substrate every other web service runs on.
Apple, Microsoft, Alphabet, Amazon, Nvidia, Meta, and Tesla. Combined market capitalisation: approximately $15 trillion as of late 2024, more than the entire UK, Japanese, French, and Indian stock markets combined. The seven companies account for roughly 30% of the S&P 500. The S&P 500's real performance over 2020-2024 is largely driven by the Magnificent Seven; the equal-weighted S&P 500 substantially underperformed the cap-weighted S&P 500. The concentration is unprecedented in US equity-market history.
Statcounter / DOJ trial evidence: Google holds approximately 90% of the US general search market and roughly 95% of the US mobile search market. The DOJ won the August 2024 ruling in United States v. Google LLC (Judge Amit Mehta) finding that Google illegally maintained its search monopoly through default-payment deals (~$26B/yr to Apple alone, plus payments to Samsung, Mozilla, etc.). The remedy phase is ongoing as of late 2024; potential remedies include forced divestiture of Chrome, banning default-payment deals, or compulsory data licensing.
Synergy Research / Gartner: AWS holds ~32% global cloud-infrastructure market share, Azure ~23%, GCP ~12%. The three firms together control roughly 65% of the global market, ~75% of the US enterprise market. Cloud infrastructure is the substrate that almost every major web service runs on, including the AI workloads of competitors. Each of the three firms is also a major AI-foundation-model developer or distributor, which creates structural conflicts: cloud customers may compete with their own cloud provider in adjacent product markets.
eMarketer / Pew Research: Meta-owned platforms (Facebook, Instagram, WhatsApp, Threads, Messenger) account for roughly 45% of US adult social-media time. YouTube ~25%. TikTok ~15%. The remaining ~15% is spread across X, Snapchat, Reddit, Pinterest, LinkedIn, and others. The 2012 Instagram acquisition ($1B) and 2014 WhatsApp acquisition ($19B) are now widely flagged as competition-killing, including in the FTC's 2020 antitrust suit against Meta. Both deals were approved at the time without meaningful competition-policy review.
Surveillance
4 itemsMeta extracts ~$300/yr in advertising revenue per US user. To produce that ARPU it tracks app usage, off-platform behaviour via Pixel, location, device fingerprinting, and inferred attributes (income, life events, political leanings). Major data brokers hold ~3,000 attributes on the average US adult and trade them in a market without comprehensive federal regulation. The EU has GDPR; Brazil has LGPD; Canada has PIPEDA; the United States has no equivalent statute and the closest near-passage attempt (APRA 2024) died in the House.
Meta Platforms 10-K: average revenue per user (ARPU) in the US/Canada region was ~$220 in 2023, ~$300 in 2024. Almost all of it advertising. To generate that ARPU, Meta tracks a range of behavioural data including app usage, off-platform website visits via Pixel and SDK integrations, location, device fingerprinting, and inferred-attribute models (predicted income, life events, political views). The "free" service is not free; the user is the product, and the price is the data extraction.
2018 disclosures (Cadwalladr / Channel 4 / Wylie): Cambridge Analytica obtained data on roughly 87 million Facebook users through a personality-quiz app (thisisyourdigitallife) that the Facebook Graph API permitted to access friends-of-quiz-takers data without their knowledge. The data was used for psychographic targeting in the Trump 2016 and Brexit campaigns. Facebook paid a $5B FTC fine (the largest in agency history at the time) and Mark Zuckerberg apologised in 2018 Senate testimony. The underlying API permissions had been removed in 2014 but Cambridge Analytica retained the harvested data.
FTC data-broker report (2014, updated 2024): major data brokers (Acxiom, Experian, Oracle Data Cloud, LiveRamp) hold roughly 3,000 distinct attributes on the average US adult. The categories: demographic, financial, health, location, purchasing, online behaviour, inferred preferences, social-graph affiliations. The data is bought and sold across the ad-tech and financial-services industries with limited US regulatory framework. The EU GDPR (2018) and California CCPA (2020) imposed disclosure and opt-out rights; federal US law has no equivalent.
Unlike the EU (GDPR 2018), Brazil (LGPD 2020), Canada (PIPEDA), the UK, India, Japan, Australia, and most other large democracies, the United States has no comprehensive federal consumer-privacy law. The American Privacy Rights Act (APRA, 2024) was the closest near-passage attempt; it died in the House. The patchwork of state laws (California CCPA/CPRA, Virginia, Colorado, Connecticut, Utah, etc.) creates fragmented coverage, particularly for inter-state data flows. The legislative gap is roughly 25 years old; HIPAA (1996) and the Children's Online Privacy Protection Act (1998) are the last major US privacy statutes.
Antitrust
4 itemsThe 1982 AT&T breakup is the last major US tech-monopoly structural enforcement. Microsoft 2001 ended in a behavioural settlement. The 25 years between Microsoft and the current Google search ruling produced the consolidation the current cases are now trying to unwind. Over the same period, the EU has imposed >€10B in tech-monopoly fines and passed the Digital Markets Act. The American Innovation and Choice Online Act died in 2022 after $120M+ in industry lobbying. The transatlantic enforcement gap is the mechanism.
The 1982 AT&T consent decree broke up the Bell System into seven Regional Bell Operating Companies (the 'Baby Bells'). The breakup was the last major US technology-sector antitrust enforcement that produced structural remedies. The 2001 Microsoft case ended in a behavioural-conduct settlement rather than the original Department of Justice request for breakup. Between 1982 and 2024, no major US tech monopoly was structurally broken up. The current Google search case is the first US antitrust ruling against a tech monopoly since 2001; the remedy phase is ongoing.
Lina Khan's 2017 Yale Law Journal note "Amazon's Antitrust Paradox" was the foundational legal argument for renewed structural antitrust enforcement against tech platforms. Khan was confirmed FTC chair in 2021 and pursued the FTC v. Meta and FTC v. Amazon cases. The Trump second term replaced her in early 2025; the Republican-controlled FTC has signalled scaled-back enforcement. The 25-year gap between Microsoft 2001 and the current cases produced the consolidation the cases are now trying to unwind. Whether the current cases will produce structural remedies, behavioural settlements, or be dropped remains contested.
EU competition rulings against US tech firms 2017-2024: Google fined €2.4B (shopping comparison, 2017), €4.34B (Android, 2018), €1.49B (AdSense, 2019); Apple fined €1.8B (music streaming, 2024); Meta fined €1.2B (data transfers, 2023). Cumulative EU fines on US tech: more than €10 billion. US Department of Justice and FTC fines on the same firms over the same period: a small fraction of that. The EU has functioned as the de-facto global tech regulator, with Brussels effects shaping US user experiences (cookie banners, GDPR notices, App Store sideloading in the EU).
The EU Digital Markets Act (in force March 2024) designates "gatekeeper" platforms (Apple, Alphabet, Amazon, Meta, Microsoft, ByteDance) and imposes structural obligations: app sideloading, third-party app stores, interoperability, ad-data sharing, user-choice screens. US Congress has not passed equivalent legislation. The American Innovation and Choice Online Act (AICOA), the leading US bipartisan tech-competition bill, died in the 117th Congress (2022) under industry lobbying pressure ($120M+ spent against it per OpenSecrets). Subsequent versions have not advanced.
Acquisitions
4 itemsInstagram cost Facebook $1B in 2012; it now generates ~$50-60B/yr. WhatsApp cost $19B in 2014 and now has ~3 billion users. Both deals were approved without serious competition review. FAANG together completed ~700 acquisitions between 2010 and 2024, almost all approved. Internal Zuckerberg emails released in FTC v. Meta describe the strategy explicitly: buy potential competitors before they grow up. Microsoft's $13B OpenAI investment was structured to avoid Hart-Scott-Rodino review entirely. The acquisition pipeline is how concentration sustains itself.
Facebook acquired Instagram for ~$1 billion in April 2012 (cash plus stock). The deal closed without serious competition-policy review at the time; Instagram had ~30M users and no revenue. Internal Facebook emails released as part of the FTC Meta antitrust suit show Mark Zuckerberg explicitly framing the acquisition as a way to "neutralize a potential competitor": "Even if some new competitors springs up, buying Instagram, Path, Foursquare, etc. now will be very expensive." Instagram now generates an estimated $50-60B per year in advertising revenue, ~50% of Meta's ad business.
Facebook acquired WhatsApp for ~$19 billion in February 2014. The deal was the largest tech acquisition in history at the time. WhatsApp had 450M users and minimal revenue; Meta's acquisition rationale was the same as Instagram: pre-empt competition. WhatsApp now has ~3 billion users globally. The 2014 EU competition review accepted Facebook's representation that user-account merging was technically infeasible; Facebook merged WhatsApp accounts into Meta's ad-targeting graph in 2016, which the EU subsequently fined ~€110M for in 2017. The acquisition was approved.
CB Insights / FTC merger reviews: Apple, Amazon, Meta, Google, and Microsoft together completed approximately 700 acquisitions between 2010 and 2024. The vast majority were approved without serious antitrust review under the Hart-Scott-Rodino threshold (~$110M as of 2024) or via the FTC's "early termination" disposition. Many small acquisitions were of competitors or potential competitors in adjacent markets (called "killer acquisitions" in antitrust scholarship). The cumulative effect has been to consolidate competing technology, talent, and intellectual property into the same five firms.
Microsoft has invested ~$13 billion in OpenAI starting 2019, structured as cloud-credit investment + equity-equivalent rights. The structure was deliberately designed to avoid triggering Hart-Scott-Rodino merger review (Microsoft does not hold a controlling stake on paper). Functionally, Microsoft has commercial integration with OpenAI's models (Azure exclusive, Copilot integration) and significant board influence. The FTC opened an inquiry in 2024; the EU Commission also opened review. The 'investment, not acquisition' structure has become a template; Amazon's Anthropic investment ($4B+) and Google's mirror it. Antitrust review framework predates this category.
Platform tax
4 itemsApple App Store and Google Play take 30% of in-app revenue. Amazon Marketplace third-party sellers pay ~50% of revenue back to Amazon across combined fees. Google pays Apple ~$26B/yr to remain Safari's default search, the single largest line-item between any two technology companies. The cumulative platform tax on the digital economy runs in the hundreds of billions per year. The pattern: own the choke point, charge rent. The 2022 antitrust legislation that would have prohibited self-preferencing died after $120M of industry opposition.
Apple and Google charge developers 30% of in-app purchase revenue and subscription revenue (with various small-developer reductions and second-year subscription rates of 15%). The structure has been challenged by Epic Games (Epic v. Apple, 2021), in the EU under DMA, and by South Korea's in-app purchase law. The 2021 ruling required Apple to permit external-payment links but largely upheld the commission structure for in-app payments. Apple's App Store generates ~$24B per year in commission revenue; Google Play ~$15B. The aggregate platform tax on the mobile app economy exceeds $40B per year.
Institute for Local Self-Reliance / GAO analyses: Amazon Marketplace third-party sellers pay an average of ~50% of revenue back to Amazon across various combined fees (referral fees ~15%, FBA fulfilment ~15%, advertising-on-Amazon ~10%, plus storage and other charges). Amazon itself competes with sellers on the platform; FTC v. Amazon (2023) alleges that Amazon uses seller data to launch competing private-label products and that Amazon punishes sellers who list cheaper prices on competing platforms. The "Buy Box" algorithm controls ~85% of marketplace sales and is opaque to sellers.
United States v. Google LLC trial evidence: Google paid Apple ~$20 billion in 2022, projected to reach ~$26 billion by 2024, to remain the default search engine on Safari and other Apple products. The payment is the single largest line item between any two private companies in technology. The DOJ's antitrust theory: the default-payment system foreclosed competition because being the default is the dispositive factor in search-market share, and only Google could afford to bid. Judge Mehta agreed. The remedy phase will determine whether the default-payment system can continue.
OpenSecrets: tech-industry lobbying against the American Innovation and Choice Online Act exceeded $120M in 2021-22, the most-spent-against bill in modern lobbying history. Apple, Amazon, Google, and Meta were the largest spenders. The bill would have prohibited self-preferencing by gatekeeper platforms (Amazon's Buy Box, Google's results, Apple's App Store ranking). It died in the 117th Congress without reaching a Senate floor vote. Subsequent reintroductions have not progressed. The cumulative pattern: comprehensive US tech-competition reform has not passed Congress in 25+ years; industry lobbying scale is the mechanism.
Where to go from here. The citizens united page documents the campaign-finance architecture that lets $120M of industry lobbying kill a competition bill. The ai page documents the next consolidation cycle as the same five firms become the substrate for foundation-model compute. The inequality page documents the wealth concentration the equity-market tier of these companies has produced.