The Trade Desk is hitching its growth to artificial intelligence with Kokai, a deep-learning ad buying copilot that already processes 13 million impressions per second—but the same AI reshaping its platform is also eroding the open web it depends on. The tension is unmistakable: can smarter automation in programmatic, CTV, and retail media offset the structural supply loss that AI‑powered search and summarized discovery are imposing on the open internet?

Kokai, unveiled in June 2023, represents the company’s most ambitious AI bet. It distributes deep learning across every layer of the buying stack—from impression scoring and bid optimization to budget pacing and dynamic audience creation—and management claims rapid client adoption. By early 2025, roughly two‑thirds of customers had migrated, with full migration targeted by year‑end. Case studies suggest measurable performance gains, but independent verification remains limited. The question for advertisers and investors isn’t whether AI can improve targeting; it’s whether that improvement outruns a bigger, slower‑moving disaster: the disappearance of the ad impressions themselves.

The Kokai Gamble: What It Is and How It Works

The Trade Desk markets Kokai as a “copilot” for media buyers. Advertisers set goals and guardrails, and the distributed architecture takes over micro‑decisions in fractions of a second, evaluating thousands of signals per impression. The company says the system handles the equivalent of more than 13 million ad impressions per second, powering features such as Koa Audiences (dynamic audience refinement), the TV Quality Index (transparency in connected TV), and the Retail Sales Index (measurement in retail media). These tools are crucial because CTV and retail media have long been opaque, with advertisers hungry for auditable outcomes.

On paper, Kokai’s engineering is a genuine feat. Real‑time scoring at that scale demands low‑latency inference, robust feature pipelines, and airtight privacy boundaries. Early adopter spend data indicates that clients on Kokai widen their budgets faster and see performance uplifts, though precise figures are company‑sourced and anonymized. Advertisers would be wise to run controlled holdout tests before shifting majority spend—reported lifts may reflect selection effects or reallocation within existing audiences rather than true incremental reach.

The initiative also extends into creative. Through partnerships with Rembrand (virtual product placements), Spaceback, and Bunny Studio, Kokai can automate dynamic ad generation across formats and languages. If these integrations scale, Kokai morphs from a pure optimization engine into a consolidated creative‑plus‑media platform, capturing more of the advertiser workflow and raising switching costs. Execution, however, depends on third‑party partners, content rights, and rigorous brand‑safety controls—areas where a misstep could slow adoption.

The AI Threat: Zero‑Click Search and Inventory Compression

While The Trade Desk weaves AI into its buying stack, a separate AI revolution is quietly shrinking its addressable market. Generative search overviews, voice assistants, and chatbot interfaces are reshaping discovery. When a user’s question is answered directly inside an AI interface, the click‑through to publisher pages plummets. Industry data shows that sessions where a generative overview appears produce link clicks in roughly 8% of cases, about half the historical average. For publishers that depend on search referrals, the traffic falloff is real: CNN’s web traffic dropped around 30% year over year recently, while Business Insider and HuffPost saw declines closer to 40%.

Fewer clicks translate into fewer ad‑supported impressions on the open web—the very inventory that independent demand‑side platforms like The Trade Desk have traditionally aggregated. As supply compresses, competition for quality impressions intensifies, potentially driving up CPMs without a corresponding boost in outcomes. More ominously, the platforms that control AI discovery are beginning to monetize those surfaces natively. Google has announced plans to place ads inside its AI Overviews; Microsoft is testing formats in Copilot. When advertiser budgets can be placed directly into AI‑powered discovery environments with minimal friction, demand consolidates inside those walled gardens, squeezing independent DSPs from both the supply and demand sides.

It’s important not to overstate the immediacy: a drop in referral clicks doesn’t erase all ad impressions. Many publishers retain direct audiences via apps, subscriptions, and social distribution. But the mix changes. Campaigns that relied on scale across long‑tail, search‑dependent sites lose reach and efficiency. Advertisers may respond by bidding more aggressively for remaining high‑quality open impressions—inflating CPMs—or by reallocating budgets to platform‑owned ad surfaces. Both outcomes erode the independent DSP value proposition.

Strategic Dodge: CTV, Retail Media, and Curated Supply

The Trade Desk is not standing still. It is steering aggressively into channels where AI search disruption has limited influence. Connected TV audiences are built through viewing behavior and platform ecosystems, not search referrals. Retail media placements are driven by point‑of‑sale data and in‑store discovery. Both are growing fast and are less exposed to the zero‑click trend. The company’s Retail Sales Index and TV Quality Index are designed specifically to measure these environments, giving advertisers confidence that their dollars are working.

Curated supply initiatives—OpenPath and the Sellers & Publishers 500+ list—aim to route buyers toward premium, brand‑safe inventory that will retain value even if general web traffic declines. By emphasizing transparency and measurement, The Trade Desk argues that advertisers will prefer an independent DSP over opaque platforms where bid dynamics are hidden. That pitch resonates when trust is scarce, though it hasn’t been tested at scale in a world where the very idea of an “open web” is shrinking.

The generative creative integrations also play a defensive role. By capturing more of the ad‑creation workflow, The Trade Desk raises its per‑client revenue and makes it harder for customers to defect. Virtual product placement and automated localization expand the total addressable market beyond media buying into creative production—a meaningful diversification if executed well.

Advertiser Playbook: Testing, Measurement, and Governance

Advertisers should approach Kokai and similar automation tools with disciplined skepticism. The technology is powerful but not infallible, and its reported lifts demand independent validation. Marketers should:
- Run controlled A/B experiments and holdout tests before migrating full budgets. Compare Kokai‑optimized campaigns against manual or legacy strategies to isolate genuine lift.
- Prioritize multi‑channel measurement frameworks that include incrementality testing, not just last‑click attribution. Holdouts help determine whether reported uplifts represent true acquisition or reallocation within existing audiences.
- Preserve a flexible creative‑plus‑media workflow that can pivot between open‑web, CTV, retail media, and platform‑native formats. Generative creative accelerates scale, but human review and brand governance remain essential.
- Monitor inventory quality and CPM trends monthly. If costs for curated open‑web placements spike without commensurate transparency or outcomes, rebalancing toward less exposed channels becomes urgent.

Investor Lens: Scenarios and Key Metrics

For investors, AI is a double‑edged sword. The Trade Desk’s long‑term narrative hinges on two simultaneous outcomes: Kokai delivers sustained, independently verifiable performance improvements, and the revenue mix shifts fast enough toward CTV and retail media to offset open‑web headwinds.

Quarterly indicators to watch include:
- Kokai spend share and adoption rate. Does migration continue apace? Do Kokai adopters expand budgets faster than non‑adopters?
- Open‑web CPM and supply trends. Structural inflation without proportional performance gains signals stress.
- The pace of AI‑discovery monetization by Google, Microsoft, and others. Rapid, ROI‑attractive ad insertion inside AI interfaces favors the platforms.
- Client churn and agency feedback. Operational friction around automation can slow migration or spark defections.
- CTV and retail media growth rates. Incremental spend in these areas must meaningfully offset open‑web decline.

Three scenarios capture the range of possible outcomes:
- Bull case (AI as accelerant): Kokai becomes the industry‑standard copilot, delivering measurable ROAS gains that deepen client stickiness. The Trade Desk captures disproportionate share in CTV and retail media while its creative‑plus‑buying stack lifts ARPU. Growth accelerates.
- Balanced case (partial threat, product offsets): Open‑web inventory compresses, but premium curated supply, CTV, and retail media offset enough budget to maintain positive revenue growth, albeit at slimmer margins. Kokai adoption stabilizes churn.
- Bear case (demand consolidates in platforms): AI‑discovery surfaces monetize rapidly with attractive ROI, drawing significant budgets away from independent DSPs. Open‑web supply shrinks materially, and CPM dynamics or client reallocation compresses The Trade Desk’s addressable market, squeezing growth.

Regulatory Wildcards

Antitrust scrutiny, content‑licensing disputes, and evolving rules on model training data could reshape the economics of AI search and digital advertising. If regulators force disclosures or alter platform monetization models, the balance between walled gardens and independent DSPs could shift in unpredictable ways. These are non‑linear risks that any investor must factor into long‑term assumptions.

The Verdict

The Trade Desk’s response to AI is both coherent and defensible: build a superior product, embed AI as a core differentiator, and pivot revenue toward channels insulated from search disruption. Kokai and the curated‑supply play are meaningful competitive assets. Their ultimate value, however, will be determined not just by technology but by advertiser behavior, CPM dynamics, and how quickly integrated AI discovery experiences are monetized.

Over the next 12–24 months, the balance between Kokai adoption and AI‑search monetization will decide whether AI acts as an accelerant or a headwind. Advertisers and investors alike should watch the measurable, high‑signal variables—spend migration, supply compression, and the pace of ad insertion inside generative AI interfaces. The Trade Desk has placed an intelligent bet. Now the market must show whether it pays off.