OpenProgrammatic Infrastructurethrough 2028 (directional to 2030)
35–45%
Most likely outcome
Curation eats the stack (sell-side ascendancy)
Programmatic CTV's DSPs and SSPs — and where the margin really sits for a platform-agnostic intermediary
Read the full research →01The Outcomes — 4 scenarios, priced by the research
Curation eats the stack (sell-side ascendancy)35–45%
Curation and sell-side decisioning become the default CTV transaction mode; SSPs (Index, Magnite, PubMatic) capture the activation-and-data margin; DSPs commoditized except for scale (TTD) or exclusive supply (Amazon). The intermediary thrives only if it becomes the curator on a neutral rail.
Drivers
- disintermediation / buy-sell collapse
- curation boom
- signal loss favoring sell-side data
- CTV deal-centricity
Signals to watch
- curated-deal-ID share crossing 80%
- a holdco standardizing on sell-side curation
- SSP CTV revenue outgrowing DSP revenue
Horizon · defining by 2027
The pincer (margin compression from both sides)30–40%
The undifferentiated spread becomes legible (LLD reconciliation, principal-buying disclosure, client audits, Amazon's low fee floor) WHILE the agnostic seat is squeezed from both sides: SSP curation absorbs activation from below, and holdco principal media (GroupM >$1B, Omnicom >$2B in 2025) plus brand in-housing absorb budget from above. The differentiated take rate survives as a shrinking, two-front niche; the limit is that in-housing is broad-but-shallow and principal media carries a ~90%-distrust wound (gap 10).
Drivers
- transparency tooling / LLD
- principal-buying disclosure
- Amazon price-floor effect
- SSP curation from below
- holdco principal media + in-housing from above
Signals to watch
- ANA transaction costs below ~20%
- a curation-fee disclosure standard
- a holdco mandating LLD-reconciled buying
- a major brand publicizing an in-housing win or retreat
Horizon · continuous through 2028
Walled-garden gravity (bifurcation, narrowed)15–25%
Amazon (Prime Video, the authenticated Roku reach) and the streamers' own stacks pull premium CTV into closed/PG environments the intermediary can place into but not capture spread within. Narrowed after gap 09: the Amazon-Roku exclusive binds only the OS-level authenticated IDENTITY layer, not all Roku inventory (Roku stays DSP-interoperable and sells direct), so a hard bifurcation is less likely than first thought; retail-media gravity and PG lock-in keep it live but bounded.
Drivers
- walled gardens opening-but-locking
- Amazon-Roku authenticated-identity exclusive
- retail-media gravity and O&O self-preferencing
Signals to watch
- Amazon opening or extending the Roku exclusive
- streamers tightening PG/ACR access
- retail-media CTV share climbing
Horizon · already partly here, bounded through 2028
Agentic routing commoditized10–20%
Agentic standards (AdCP/AAMP) and platform agents mature enough that advertisers' own agents route path-agnostically from a prompt, collapsing the agnostic-routing service into the protocol layer; the intermediary's optionality edge erodes unless it owns proprietary data or agents.
Drivers
- agentic AI
- AAMP/AdCP standards
- automation of trading labor
Signals to watch
- AAMP advertiser-side agents in production at a major brand
- PubMatic 25%-by-2028 autonomy forecast tracking
- a holdco deploying its own buyer agent
Horizon · emerging through 2028, full force 2028-2030
Probabilities are the research's own scenario bands, priced as outcomes. The full argument — history, current state, drivers, and sources — lives in the corpus: read the Future Trajectory chapter.
