Ad Tech Explained: DSPs, SSPs, and Why the Open Web Is a Structural Loser

Executive Summary

Digital advertising is a $700B+ global market projected to reach $900B by 2027. But the distribution of value within that market is deeply uneven — and increasingly concentrated in walled gardens (Google, Meta, Amazon) at the expense of the open web. For investors, the critical analytical task is distinguishing between companies structurally positioned to benefit from ad market growth and companies that are structurally disadvantaged by the same dynamics that drive walled garden dominance: first-party data, closed-loop measurement, and audience scale. This primer maps the ad tech stack, explains why the open web faces structural headwinds, and identifies where value is actually being created.

The Ad Tech Stack Explained

The programmatic advertising supply chain involves multiple intermediaries between the advertiser and the publisher's ad slot:

Advertiser side:

• Agency Trading Desk (ATD): Large agencies (WPP, Omnicom, Publicis) operate in-house programmatic buying desks that use DSPs on behalf of clients.
• Demand-Side Platform (DSP): Software that lets advertisers bid on ad inventory in real time. The Trade Desk, Google's DV360, Amazon DSP, and Xandr (Microsoft) are the leading DSPs.
• Data Management Platform (DMP): Aggregates third-party audience data for targeting. Largely obsolete as cookie deprecation eliminates the third-party data that DMPs depended on.

Supply side:

• Supply-Side Platform (SSP): Software that lets publishers manage and auction their ad inventory. Magnite, PubMatic, and Google's Ad Manager are the leading SSPs.
• Ad Exchange: The marketplace where DSP bids meet SSP supply in real-time auctions (auction duration: ~100 milliseconds). Google's AdX, OpenX, and Index Exchange operate exchanges.

The flow of a programmatic transaction: A user loads a web page → the publisher's SSP calls ad exchanges → exchanges send bid requests to DSPs → advertisers' DSPs evaluate the user against targeting criteria and submit bids → winning bid's creative is served → the entire process completes before the page loads (~150ms).

The take-rate problem: Each intermediary extracts a fee. DSPs typically charge 15–20% of media spend; SSPs charge 10–15% of publisher revenue; exchanges take 5–10%. In a worst-case scenario, a $1.00 advertiser spend becomes $0.40–0.50 in publisher revenue after middlemen. This "ad tech tax" is a persistent source of controversy and the core reason direct-to-publisher deals (Spotify, New York Times) command premiums.

The Walled Gardens: Google, Meta, Amazon — Why They Win

Walled gardens control their entire ad serving stack internally and do not participate in open programmatic auctions for their premium inventory:

Google ($280B+ ad revenue, 2025): Google Search (intent-based), YouTube (video), Display/GDN (open web), and Gmail/Discover (owned properties). Google's first-party data advantage — signed-in users across Search, Maps, Gmail, YouTube — allows audience targeting with no reliance on third-party cookies. Google's measurement superiority (attributing offline purchases to online ads via Google Pay and Google Wallet data) is its single largest competitive advantage over the open web.

Meta ($155B+ ad revenue, 2025): Facebook + Instagram + WhatsApp + Threads. Meta's social graph creates the richest behavioral targeting dataset in existence — relationships, interests, life events, employer, political views (inferred). Meta's Advantage+ AI bidding system automates campaign optimization across its network, consistently outperforming manual bidding. The iOS 14.5 ATT framework (2021) cost Meta ~$12B in lost 2022 revenue, but Meta's response — building the Conversions API for server-side tracking and investing in modeled attribution — has substantially recovered measurement capability.

Amazon ($55B+ ad revenue, 2025): Amazon's ad business is the fastest-growing of the three, driven by sponsored products within search results (a direct Google Search analog for product queries). Amazon's data advantage is purchase behavior — knowing what consumers actually buy is more valuable than knowing what they search. Amazon DSP extends this targeting off-Amazon to the open web, making Amazon increasingly competitive with Google for lower-funnel e-commerce campaigns.

Why walled gardens structurally win:

• First-party identity (logged-in users) survives cookie deprecation completely.
• Closed-loop measurement (ad impression → purchase in the same ecosystem) outperforms open web attribution models.
• Audience scale that open-web publishers cannot match individually.
• No intermediary take-rate — every dollar goes directly to the platform.

The Open Web: Structural Headwinds

The open web — news publishers, sports sites, recipe blogs, review platforms — monetizes almost entirely through programmatic advertising. This model faces four simultaneous headwinds:

1. Cookie deprecation: Google's deprecation of third-party cookies in Chrome (phased in 2024–2025, substantially complete by 2026) eliminates the user tracking layer that powered open web targeting. Publishers that relied on behavioral targeting segments from third-party data vendors face CPM compression of 30–60% on affected inventory, per industry estimates.

2. Signal loss and measurement degradation: Even before cookie deprecation, iOS 14.5's ATT framework (2021) eliminated IDFA-based tracking on iOS. Combined with Firefox and Safari blocking third-party cookies by default since 2019–2020, approximately 50% of open web users are already cookieless. The remaining Google Chrome users (65% of browser share) represent the final cookie deprecation event.

3. Margin compression from intermediaries: The average publisher earns $0.40–0.55 per $1.00 of advertiser spend after SSP, DSP, exchange, and verification tool fees. As CPMs compress, the fixed-cost intermediary layer consumes an increasing share of a declining total.

4. Audience fragmentation: Open web publishers individually lack the scale to be meaningful for national advertisers. A food blog with 5 million monthly visitors is trivial relative to Instagram's 2 billion daily actives. Aggregation (through SSPs and PMPs — private marketplaces) partially addresses this, but the structural disadvantage persists.

The Trade Desk: Best Positioned Open Web Survivor

The Trade Desk (TTD, $40B+ market cap, 2026) is the largest independent DSP, processing ~$11B in annual media spend. Its investment thesis rests on a critical structural position: TTD does not own media inventory (unlike Google's DV360, which conflicts with Google's publisher interests), so it has a credible claim to neutrality.

Why TTD is different:

• No inventory conflict: TTD routes spend to the best available inventory across 100+ SSPs. It has no financial incentive to preference any publisher.
• Unified ID 2.0 (UID2): TTD's open-source identity framework — built on hashed, encrypted email addresses — is the most widely adopted post-cookie identity solution in the industry. 800+ publishers have adopted UID2; Publicis, IPG, and Omnicom have integrated it.
• CTV leadership: TTD processes ~45% of all programmatic CTV (connected TV) spending — the fastest-growing ad format in the open ecosystem.
• Retail media data partnerships: TTD's Galileo platform integrates first-party retail data from Walmart, Target, Home Depot, and others, allowing targeting based on purchase behavior on the open web — partially replicating Amazon's data advantage.

Risks: TTD is dependent on the open web's health. If walled garden share continues to expand from its current ~75% of digital ad spend, TTD's addressable market compresses.

Retail Media Networks: Walmart Connect, Amazon DSP, Kroger

Retail media — ads served within or adjacent to a retailer's owned properties (website, app, store kiosks) — is the fastest-growing advertising sub-category at $125B+ globally in 2025.

Retail media's structural advantage is closed-loop measurement: an ad served on Walmart's website can be directly tied to a purchase at Walmart, with no attribution modeling required. This measurement superiority justifies CPM premiums of 50–100% over open web inventory.

CTV Advertising: Where Growth Is Concentrated

Connected TV — streaming services viewed on smart TVs and devices — will command $30B+ in U.S. ad spend in 2026, growing 18–22% annually. The key players:

• Netflix: Launched its ad-supported tier (Basic with Ads) in 2022; 40M+ monthly ad-supported subscribers by 2025. Netflix's targeting is based on its own viewing data — a strong contextual signal.
• Amazon Prime Video: Began inserting ads into Prime Video standard tier in early 2024; ~$2B+ ad revenue in first year.
• Hulu: Disney's ad-supported streaming service; 50M+ subscribers, ~$4B ad revenue.
• Peacock, Paramount+, Max: Each building ad-supported tiers as the SVOD growth plateau forces streaming companies to diversify revenue.
• YouTube (CTV): YouTube on TV screens is the #1 streaming service by watch time in the U.S. (per Nielsen). YouTube's CTV revenue is growing 30%+ annually.

CTV benefits from television's premium CPM environment ($30–60 CPM vs. $3–8 CPM on open web display) with programmatic buying efficiency. The Trade Desk processes the largest share of non-walled-garden CTV; Magnite is the leading CTV SSP.

Identity Solutions Post-Cookie

The post-cookie identity landscape is fragmented:

• UID2 (The Trade Desk): Email-based, open-source, privacy-preserving. Most widely adopted industry solution in North America.
• Google Privacy Sandbox: Chrome's Topics API replaces behavioral targeting with browser-side interest categorization. Low adoption from buyers and sellers as of Q1 2026.
• Seller-Defined Audiences (SDA): Publisher-side contextual segmentation transmitted via OpenRTB. Growing adoption but limited targeting precision.
• First-party data activation: Publishers with owned identity data (subscribers, registered users) can activate that data directly without third-party intermediaries. New York Times, Financial Times, and ESPN are leaders in this approach.
• Clean rooms: Data clean rooms (AWS Clean Rooms, LiveRamp Clean Room, Google Ads Data Hub) allow advertisers and publishers to match first-party data without exposing raw user data. Walmart Connect and Target Roundel use clean rooms as the primary measurement infrastructure for retail media campaigns.

How to Think About Ad Tech Stocks

The ad tech sector is bifurcated between structural winners (walled gardens and open web survivors with identity solutions) and structural losers (intermediaries dependent on third-party cookies):

Structural winners:

• Alphabet (GOOGL): Search is still 57% of digital advertising. YouTube is the largest CTV platform.
• Meta (META): Social + visual intent data is irreplaceable for consumer brand advertising.
• Amazon: Purchase data is the best lower-funnel signal in existence.
• The Trade Desk (TTD): Neutral DSP with UID2 identity advantage and CTV leadership.

Structural neutrals:

• Magnite (MGNI): Leading independent CTV SSP but dependent on open web health for non-CTV revenue.
• PubMatic (PUBM): SSP with solid CTV positioning; balance sheet strong for its scale.

Structural losers:

• Legacy DMP businesses (Oracle Advertising, Lotame): Third-party data businesses with no credible post-cookie pivot.
• MFA (Made for Advertising) publishers: Low-quality open web publishers dependent on remnant programmatic at minimal CPMs.

Comparison Table: Walled Gardens vs. Open Web Players

Takeaways for Investors

• Walled gardens will take 80%+ of digital ad spend by 2027: The structural dynamics (first-party data, closed-loop measurement, audience scale) are self-reinforcing. There is no credible scenario in which the open web reverses this trend.
• The Trade Desk is the only public equity proxy for open web upside: Its neutral positioning, UID2 ecosystem, and CTV leadership make it the best-positioned non-walled-garden ad tech company. The premium valuation is justified by its structural differentiation.
• Retail media is the next $200B market: Walmart, Kroger, Target, and Instacart are building ad businesses that will collectively rival the television industry within five years.
• CTV is where television ad budgets are going: Every linear TV dollar that shifts to streaming is an addressable programmatic opportunity. The winners are YouTube (within Google), Netflix, Hulu, Amazon Prime Video, and TTD as the independent buyer platform.
• Avoid pure-play cookie-dependent businesses: Any ad tech business whose primary value proposition depends on third-party behavioral data faces structural erosion that no pivot will fully reverse.