The Business Model War Beneath the Protocol War

Strip away the technical details and you find a straightforward conflict between two economic models.

Google’s $264.6 Billion Problem

Google’s annual advertising revenue depends on being the intermediary between consumer intent and merchant fulfillment. When that intermediation happened through ten blue links, advertising could be inserted alongside organic results.

When it happens through conversational AI that synthesizes information and completes transactions, the old the intelligence factory race between AI labs — s-business-model-breaks/”>model breaks.

Google’s solution is to embed advertising into the new model:

• Direct Offers ensures advertising travels with the transaction into conversational interfaces
• UCP ensures Google remains essential infrastructure for the shopping journey
• The bet: advertising can be embedded into the moment of decision, not just the moment of discovery

OpenAI’s Evolution

OpenAI started from a different position. Without Google’s advertising legacy to protect, OpenAI experimented with transaction-fee economics—capturing a percentage of completed purchases rather than charging for visibility.

But the compute costs are brutal, and the agent revenue bet hasn’t paid off. OpenAI cut its 2025 agent revenue projections by half to $1.4B.

The math became obvious: if transaction fees can’t fund the infrastructure — as explored in the economics of AI compute infrastructure — at required scale, advertising becomes inevitable. The January 2026 advertising announcement wasn’t a strategic choice—it was a financial necessity.

Where They Converge

Despite the business model divergence, the protocols share critical architectural DNA:

• Both preserve the merchant as the “merchant of record”—neither wants to become Amazon
• Both protocols are payment processor agnostic by design
• Both are built for modular extension
• Both acknowledge MCP as foundational

This is platform power without platform liability.

The Core Economic Divergence

Both converge on ads eventually. They differ on where and how ads are inserted:

• Google: Advertising embedded inside commerce. Ads travel with the transaction
• OpenAI: Ads bolted onto the interface. Commerce and ads remain structurally separate

History suggests integrated advertising usually wins. But AI interfaces may follow different rules if users reject ads in conversations they perceive as personal assistance.

This is part of a comprehensive analysis. Read the full analysis on The Business Engineer.

How AI Is Reshaping This Business Model

AI fundamentally rewrites the intermediary economics that define the search-advertising duopoly. Traditional search requires users to manually navigate through multiple queries and results pages, generating numerous ad impressions per intent. AI assistants collapse this process into direct answers, potentially reducing Google’s ad inventory by 60-80% per user session. The War Beneath The Protocol War sits at the center of this transformation. While Google scrambles to integrate AI without cannibalizing its $264.6 billion advertising engine, pure-play AI companies like OpenAI can optimize for user satisfaction rather than engagement metrics. This creates asymmetric advantages—OpenAI can afford to give users exactly what they want immediately, while Google must balance helpfulness against revenue protection. The operational shift is equally dramatic. AI inference costs are front-loaded and predictable, unlike the variable costs of maintaining massive search indexes and advertising auction systems. Companies can now charge subscription fees for guaranteed access to intelligence, rather than harvesting attention for advertising dollars. Within five years, the protocol war’s winner will be determined not by superior algorithms, but by which business model better survives the transition from attention economy to intelligence economy.

For a deeper analysis of how AI is restructuring business models across industries, read From SaaS to AgaaS on The Business Engineer.