The Jeff Wars: Independence vs. Integration in Ad Tech

Independence isn’t a moat anymore

For years, an “independent DSP” signaled neutrality and wide-open access to supply. That advantage narrows when integrated platforms—whether Amazon, Google, Meta, or emerging retail media networks—can combine authenticated identity, purchase data, premium video inventory and pricing control in one stack. While Amazon offers one of the clearest examples (its advertising services hit roughly $15.7 billion in Q2 2025, up about 23% year over year), this same pattern is playing out across multiple walled gardens.

The buying center of gravity is shifting toward integrated platforms that combine proprietary inventory with rich first-party data—think YouTube ads with Merchant Center product feeds and audience signals, Meta’s graph plus Shops/CAPI, and Walmart Connect applying shopper data to CTV and shoppable streaming. 

By Q4 2024, the Amazon DSP accounted for ~32% of all Amazon ad spend among advertisers in one large dataset, while Google's DV360 maintains similar dominance for YouTube inventory access.

‍Price matters too. Multiple agency sources say Amazon has been undercutting rival DSPs on platform fees and offering aggressive commercial terms to win share; some buyers reported tech fees trimmed to around 10% in recent months. When you pair lower fees with access to logged-in audiences and closed-loop retail attribution, independent DSPs face a tougher sell on cost–performance grounds.

Inventory access is tilting as well. Each major platform is increasingly gating premium supply: Amazon with Prime Video ads available via Amazon DSP in both managed-service and self-serve modes; Google with YouTube Select premium lineups; Meta with Reels inventory. And the June 2025 Amazon–Roku deal makes Amazon DSP the exclusive way to reach Roku’s logged-in CTV audience via a shared identifier.

The trend is clear: platforms are scaling ad-supported viewing and tightening access. 

Independence still matters for control, transparency and cross-publisher activation, but make diversification the hero: keep an open-internet DSP and at least one platform-native route live, and adjust the mix based on performance, bandwidth and scope rather than ideology.

‍

{{the-jeff-wars-1="/tables"}}

‍

Over-reliance on a single DSP is risky

The Trade Desk’s sharp, ~38% single-day stock drop after its August 2025 earnings and CFO transition was a reminder that even category leaders can stumble. Concentrating spend with any one platform compounds operational and commercial risk—guidance resets, leadership changes or product shifts can quickly ripple into pricing, service levels or roadmap priorities.

There’s also platform continuity risk you don’t control. In May 2025, Microsoft announced it will shut down Microsoft Invest (formerly Xandr/AppNexus) by early 2026, pivoting toward AI-powered buying tied to its own properties. Buyers dependent on that DSP now face migrations, integration rewires and learning-curve costs, illustrating why vendor concentration can become a single point of failure.

Financial plumbing adds another layer. AdExchanger detailed how extended payment terms and late invoices create a cash-flow gap for DSPs that can strain even well-run platforms. If a DSP is squeezed between slow agency payments and faster obligations to supply partners, service disruptions or retrenchment become more likely. A diversified stack helps insulate campaigns from those shocks.

Exclusivity and identity access also argue against putting all spend in one tool. 

Each major platform gates certain inventory. If you're not enabled on multiple platforms, you're structurally boxed out of key audiences. The smart play is maintaining access to multiple inventory sources.

Finally, measurement and control benefit from redundancy. Senior analysts and TV buyers note that different DSPs see different supply, apply different algorithms and expose different reporting, which means a second primary DSP can act as a benchmark and a release valve when inventory, fees or performance shift. 

In short: independence on its own is no longer enough, and dependence is a liability. Treat optionality across at least two scaled buying routes as standard risk management.

Budgets are flowing where the data lives

Buyers are concentrating dollars where authenticated audiences and purchase signals are strongest. Amazon is a prominent example, but think of it as one case among many; the broader shift is toward platforms that pair logged-in reach with measurable outcomes. Its latest earnings reinforced faster-than-market momentum and confirmed heavyweight status, not a challenger.

‍

{{the-jeff-wars-2="/tables"}}

‍

Retail media keeps taking a larger slice of US budgets because it links ad exposure to closed-loop sales. Insider Intelligence/eMarketer expects US retail media to top $62 billion in 2025, adding more than $10 billion year over year; other forecasts peg Amazon’s retail-media ad revenues alone at $60+ billion in 2025 and nearly $70 billion in 2026. Importantly, growth isn’t confined to one player: Walmart Connect, Instacart Ads, Kroger Precision Marketing and Roundel are also capturing share as marketers chase authenticated audiences and verified outcomes.

CTV is the other magnet. Insider Intelligence estimates US CTV ad spend will reach ~$33.3 billion in 2025 (≈16% YoY), driven by major streamers’ ad tiers. Coming out of the 2025–26 upfronts, several leading streamers (including Amazon) confirmed higher ad loads and new formats, expanding the pool of premium, logged-in inventory available to buyers.

Most of that expansion is paid for by cuts elsewhere. Forecast downgrades and tariff-driven budget caution are prompting reallocation toward channels that prove performance quickly. Digiday’s interviews with holding-company and independent buyers indicate more clients are shifting incremental spend into platform-native routes (Amazon’s DSP among them) in 2025 to de-risk plans.

Broader market outlooks also point to retail media and performance-oriented channels holding up better than broad social or open-web display when budgets tighten.

The next differentiator: AI + agility

Closed ecosystems are entrenched, so the edge comes from how fast you can plan, activate, and measure across them. The most credible gains right now are in time-to-insight and cross-channel decisioning. For instance, Dentsu reports a 90% reduction in time to media insights after deploying an Azure-based agent that automates data prep and analysis for planners. For mid-sized teams, that speed matters even in a modest pilot—when you’re rebalancing a handful of campaigns across a platform-native route, an open-internet DSP, and direct CTV, rather than rebuilding the whole stack.

Clean rooms are the connective tissue. Amazon expanded Amazon Marketing Cloud this year to allow queries against up to five years of store purchase signals, extending beyond the prior 13-month window. Buyers are using AMC to unify path-to-purchase analysis across retail media and CTV, then piping learnings back into activation. Roku, for its part, evolved its clean-room offering into Roku Data Cloud to expose more granular streaming data via ad-tech and agency APIs—useful for deduplicating reach and frequency when you’re buying both platform-native and open-web CTV. 

You don’t need an enterprise build to start; many teams begin with a platform clean room (or lightweight joins) and graduate to broader collaboration as scope and resourcing allow.

Identity standards help you move between walled and open environments without rebuilding everything from scratch. The IAB Tech Lab’s PAIR protocol formalizes privacy-safe match-ups between advertiser and publisher data across the clean-room ecosystem, while UID2 continues to gain adoption among CTV platforms (e.g., LG, Roku partnerships) to improve addressability off-platform. Aim for an interoperable setup that uses what your partners already support, rather than chasing a single ID to solve every use case.

On the platform side, major sellers are pushing more native automation and full-funnel workflows—Amazon within its DSP, and similar automation is advancing across other large streamers and buying tools. Agencies are pairing those native capabilities with parallel learning agendas on a second DSP to benchmark algorithms, verify incrementality, and maintain control over pacing and bid strategies.

Analyst houses agree that AI’s payoff is operational: faster cycles, better targeting from first-party data, and more precise measurement. McKinsey’s latest AI research flags workflow redesign and governance as prerequisites for value capture; Gartner’s 2025 outlook reinforces that marketing leaders should treat AI as a process accelerator embedded in planning and activation, not a silver bullet. 

Built this way, even smaller teams can shift spend within days as channels or fees change—without promising enterprise-scale lifts out of the gate.

📌 What to do now: set two primary routes, keep the clean room (or a platform equivalent) as your measurement hub, and use the same audiences and a single business KPI across both. Run parallel tests where it fits, let automation shorten the path from signal to decision, and steer budget toward what proves lift. This won’t unwind consolidation, but it will give you the flexibility to work confidently inside and outside the gardens.

Where this leaves agencies and brands

The “Jeff Wars” aren’t really Jeff Green vs. Jeff Bezos; they’re independence vs. integration, optionality vs. consolidation. Agencies caught in the middle will need to decide: double down on a single platform’s promise, or build a multi-DSP strategy that ensures flexibility.

From my perspective, the winners won’t be the ones who pick the “right” DSP, but the ones who refuse to put all their chips on the table of any one provider. Independence alone isn’t enough, but independence combined with open access and the ability to move fluidly across ecosystems is what will keep agencies resilient.

In other words, the future belongs to those who operate with an open garden mindset: neutral, agile, and ready to connect the dots across walled gardens and the open internet alike.

If you’re rethinking your plan, we can help. AI Digital works with brands and agencies to pressure-test their mix, map risk exposure, and stand up a two-DSP baseline. If you’d like a working session to explore the best route for your team, reach out and we’ll chart a practical plan together.