On August 29, 2025, People Inc. filed an 88-page antitrust complaint against Google in the Southern District of New York. The filing alleges that Google's ad tech monopoly — 90% of the publisher ad server market, 60–70% of the ad exchange market — systematically depressed what publishers earn from their own inventory.
This came four months after Judge Brinkema ruled that Google violated federal antitrust law by monopolizing the publisher ad server and ad exchange. The DOJ is pushing for forced divestiture — the first Big Tech breakup order in decades.
People Inc is suing the company that runs its ad infrastructure. If you work in their engineering org, this is the most consequential architecture problem you aren't planning for yet.
The Closed Loop
Google's ad tech monopoly isn't a market position. It's an architecture. Understanding why it's so hard to escape requires seeing how the pieces fit together.
People Inc — like nearly every large publisher — runs its programmatic advertising through this loop. Their sites use Google Ad Manager. Their exchange demand flows through AdX via Exchange Bidding (Google's server-side alternative to header bidding). Their ad decisioning happens inside Google's walls.
D/Cipher is the intelligence layer on top of this plumbing — contextual targeting that tells advertisers what a page means. But the pipes that carry the money? Those are Google's.
The Secret Projects
People Inc's 88-page filing, and the DOJ's evidence in the underlying case, detail specific mechanisms by which Google extracted value from publishers. These aren't allegations of vague "market power." They are documented engineering projects with codenames.
| Project | Year | Mechanism | Publisher Impact |
|---|---|---|---|
| Bernanke | 2013 | Manipulated bid prices in AdX auctions. Google Ads submitted bids above advertiser budgets to win impressions. | Up to 40% revenue reduction |
| Last Look | — | AdX could view competitor bids before submitting its own. Won by marginal amounts. | Systematic underpayment |
| Poirot | 2017 | Reduced DV360 bids to rival exchanges by 10–90% to kill header bidding competition. | $200M shifted to Google; rival exchanges −20–30% revenue |
| Jedi Blue | 2018 | Deal with Meta to undercut header bidding. Facebook got preferential auction terms in exchange for reduced competition. | Header bidding adoption suppressed |
| Dynamic Revenue Share | — | Google selectively adjusted exchange commissions based on competitor bidding patterns. | Unpredictable yield |
People Inc's filing claims these mechanisms collapsed the premium differential between direct-sold and programmatic inventory by roughly 50%. For a publisher doing hundreds of millions in digital ad revenue, that's not a rounding error. It's a different business.
Google's internal documents, surfaced during the DOJ trial, acknowledged Project Bernanke was "inherently unfair" and operated under strict secrecy. The system generated approximately $30 billion annually for Google.
Two Kinds of Google Dependency
Here's what makes People Inc's position architecturally unusual. They've spent three years building targeting independence from Google through D/Cipher — contextual signals that replace cookies. But they haven't built infrastructure independence. The targeting layer and the plumbing layer are different problems.
D/Cipher+: Extends contextual targeting to third-party inventory via open web.
Status: Operational. Over 50% of direct digital revenue uses D/Cipher. Expanding off-platform.
Exchange: AdX via Exchange Bidding (server-side). Demand flows through Google's auction.
Status: Fully dependent. Migrating is "rewiring the engine mid-flight" — a year of disruption, per industry estimates.
D/Cipher is the intelligence layer. GAM + AdX is the plumbing layer. People Inc built a brilliant contextual brain that still runs on monopoly pipes.
Three Remedy Scenarios
Judge Brinkema is currently writing her remedies opinion, expected mid-2026. The DOJ and Google have proposed very different futures. Each creates a different engineering problem.
Scenario 1: Forced Divestiture
What happens: Google must sell GAM and/or AdX to a qualified buyer. The ad server and exchange become separate companies.
Engineering impact: The buyer inherits a monopoly product with declining investment incentive. GAM's integration with Google's demand stack breaks. Exchange Bidding stops working as designed. Publishers need to evaluate whether the divested GAM is still the best ad server — or whether this is the window to migrate to Prebid Server and a multi-SSP setup.
Timeline: 2027–2028 at earliest, after appeals.
Scenario 2: Behavioral Remedies
What happens: Google keeps its products but faces restrictions — must allow publishers to use AdX without GAM, must share auction data, must not give AdX preferential treatment in GAM.
Engineering impact: Smallest disruption. Publishers gain the option to diversify without being forced to. The smart move becomes building an abstraction layer that can route between GAM and alternatives without rebuilding the entire stack.
Timeline: Could take effect within months of the ruling.
Scenario 3: Google's Prebid Bridge
What Google proposed: Technical integrations that let publishers receive AdX demand via third-party ad servers and Prebid in real-time.
Engineering impact: Google keeps everything but opens API access. Sounds good until you remember the last time Google "opened" its stack — Exchange Bidding was the alternative to header bidding that kept everything server-side inside Google's walls.
Timeline: Could ship faster, but trust deficit is the actual constraint.
What to Build Now
The remedy is coming. The timeline is uncertain. But the engineering work isn't scenario-dependent — every outcome rewards the same preparation.
Build the abstraction layer between your ad logic and Google's plumbing. D/Cipher already provides targeting independence. The missing piece is auction independence — the ability to route demand through GAM, Prebid, Magnite, Index Exchange, or whatever the post-remedy landscape looks like, without a year-long migration.
Three concrete priorities:
1. Prebid adapter readiness. Prebid.js is the open-source standard for header bidding. Even if you keep GAM as primary, having a tested Prebid integration means you can add SSP competition in weeks, not months. DDM currently uses Exchange Bidding (server-side, inside Google's walls). Building the Prebid path gives you the exit door.
2. Server-side bidding diversity. Don't wait for divestiture to test non-Google demand. Run A/B tests with Magnite, PubMatic, and Index Exchange on a small percentage of traffic. Measure the actual revenue delta. The lawsuit claims 50% premium erosion — the only way to know what your inventory is actually worth is to let other exchanges bid on it.
3. D/Cipher as the portable layer. D/Cipher's contextual signals shouldn't be locked to GAM's targeting keys. If the contextual taxonomy and intent data can be passed to any ad server or SSP as standardized key-value pairs, D/Cipher becomes infrastructure-agnostic. This is already partially true with D/Cipher+'s off-platform expansion — extend the same principle to on-platform ad serving.
People Inc is simultaneously suing Google for anticompetitive extraction and running its entire ad infrastructure on Google's stack. That's not hypocrisy — it's an accurate description of what monopoly lock-in looks like. The 88-page filing is the diagnosis. The engineering work above is the treatment. The remedy, whenever it arrives, is just the permission slip to leave.
Sources
AdMonsters — People Inc's Antitrust Lawsuit · AdExchanger — Google's Secret Projects · Digiday — Remedy Phase Explained · AdExchanger — Imagining GAM Divestiture · Digiday — Remedies vs Reality · TechPolicy.Press — How Google Manipulated Ad Prices