From Ads to Infrastructure: OpenAI Challenges Google’s Dominance in the IT Revenue Model

Attempts to Overcome the Limits of Ad-Dependent Systems Multiply
Moves to Redefine the Information Market Beyond the Search Bar
Uncertainty Over the Viability of “Non-Ad” Business Models

The race for dominance in the artificial intelligence (AI) operating system landscape is now a head-on battle between Google and OpenAI. While Google remains cautious about fully integrating AI into its ad-centered ecosystem, OpenAI is seeking to build a “personalized ecosystem” based on subscriptions, assistants, and search tools rather than ad placements. With its “Agent Kit” and “ChatGPT Atlas,” OpenAI is accelerating expansion beyond traditional software boundaries, though infrastructure dependence and mounting deficits remain key obstacles. Industry analysts predict that the extent to which OpenAI’s experiment can disrupt the advertising-driven market structure will determine the next shift in industry leadership.

Ad-Based Ecosystem vs. Infrastructure and Subscription Models

According to IT outlet Digitimes on the 3rd (local time), the AI sector has effectively narrowed into a contest between Google and OpenAI, and whichever firm resolves its strategic challenges more efficiently will likely seize market leadership. The outlet noted that “Google possesses vast technical infrastructure but remains confined to its decades-old ad-based model, preventing full-scale AI integration.” As for OpenAI, it observed that “despite rapid growth through innovation and user expansion, its lack of proprietary computing infrastructure and massive research and development (R&D) costs have heightened financial strain.”

The two firms’ rivalry stems from fundamentally different structures. Google boasts advanced internal resources—core algorithms such as Transformer, its TPU processors, and DeepMind—but worries that reduced ad exposure could damage its profit base, leading to a cautious rollout of AI integration. OpenAI, on the other hand, has gained agility through fast experimentation and user growth but faces growing pressure due to its dependence on Microsoft (MS) Azure’s cloud infrastructure. Analysts estimate OpenAI’s operating loss could reach up to USD 7.8 billion this year.

OpenAI’s chosen breakthrough is the “Agent Kit.” Unveiled during the company’s annual developer conference “DevDay 2025,” the tool enables users to visually assemble message sorting, data linking, widget configuration, and privacy and safety controls without writing code—and deploy finished agents to websites within minutes. The demonstration showcased a scheduling assistant, including request/response logging, performance checks, and anonymization filters. OpenAI said, “The core goal is to help both individual developers and companies turn ideas into products quickly.”

Through this, OpenAI aims to evolve ChatGPT from a “system that answers anything” to one that “can be instructed to do anything.” The strategic purpose lies in expanding monetization channels and partner reach. Once agents are integrated into real workflows, conversion paths through subscriptions, APIs, and enterprise tools multiply, while conversational interactions shift toward actionable automation—raising engagement, revisit rates, and payment willingness. Accordingly, three factors—ecosystem expansion speed, user profitability metrics, and partnership accumulation—will serve as benchmarks for OpenAI’s transition to profitability.

Improved Accessibility but Persistent Gaps in Accuracy and Source Verification

OpenAI is also venturing into the search market, pushing conversation-based exploration to the forefront. Released on the 21st of last month, the AI-powered browser “ChatGPT Atlas” allows users to describe their intent conversationally instead of typing keywords, after which the browser autonomously searches the web and summarizes results. For example, when browsing vacation options, one might say, “Compare similar flight and hotel prices,” and the system compiles and presents results as recommendations or summaries—eliminating the need for address-bar input, tab switching, or new window navigation.

The company intends to redefine the browser as a personalized information assistant by replacing the traditional “keyword–click–link” flow with “conversation–auto search–summarized result.” OpenAI explained, “Instead of clicking through listed links, users can compare options, adjust parameters, and conduct follow-up searches within the same conversation window.” Early reactions have been positive, though observers note that the coexistence of familiar address bars, shortcuts, and tab management warrants further observation.

Market impact can be gauged by shifts in share and user behavior. Statistics show that as of July, 5.99 percent of desktop browser searches were conducted through large language models (LLMs)—roughly double the figure from a year earlier. As this share grows, competition among Big Tech firms and AI startups is intensifying. Google introduced “Gemini in Chrome” to deepen AI integration, while Perplexity launched “Comet” and The Browser Company rolled out “Dia,” entering the conversational search race. This signals a fundamental shift in how users access data.

However, the shift toward “conversation as search” does not immediately resolve concerns over accuracy and reliability. Conversational responses tend to simplify sources and context, prompting calls for integrated source labels, original-view options, and verification tools. Even so, analysts view the change as meaningful: browsers now detect user intent, summarize data, and automate subsequent tasks, moving the starting point of information access from keyboard input to conversation—enhancing both convenience and productivity.

OpenAI’s Profitability on Trial

Whether these emerging models can truly transform the ad-revenue foundation of the IT ecosystem remains uncertain. The key issue is whether conversational search can replace the click- and impression-based logic of the current advertising economy. When summarized responses replace traditional search result pages, ad slots shrink, and pricing must shift from impressions to task completion or verified outcomes. Google, wary of disrupting its existing framework, is proceeding cautiously, while OpenAI has positioned itself at the vanguard of change. Analysts warn that, at least in the short term, the overlap between both models could create a zero-sum battle over traffic and revenue.

Opinions on OpenAI’s financial outlook remain divided. CEO Sam Altman maintains that there are no immediate plans for an initial public offering (IPO), yet optimists predict the company could eventually raise up to USD 60 billion through one. Supporters point out that OpenAI and Microsoft recently revised their investment terms, expanding Azure cloud commitments to USD 250 billion—suggesting that preemptive spending on R&D, data centers, and chip procurement could soon yield tangible returns.

Skeptics, however, highlight the fragility of OpenAI’s cost structure. The company’s annual AI training and inference expenses are estimated at USD 7 billion, with labor costs around USD 1.5 billion, while revenue-sharing agreements (up to 20 percent of cash flow) hinder profitability improvements. Moreover, the mismatch between subscription pricing and usage, along with the slow decline of inference costs, poses risks. Reuters reported that “ChatGPT currently has about 8 million users, but only 5 percent are paid subscribers.”

As a result, debate over OpenAI’s valuation continues to intensify. Bloomberg noted that “the company’s valuation has reached USD 500 billion, rekindling concerns of an AI bubble,” adding that “it remains uncertain whether heavy investments will translate into real profits.” Even Altman himself acknowledged, “I agree that investors are overly excited.” In this light, while OpenAI’s experiment may have the potential to disrupt ad-based models, its high fixed costs, revenue sharing, and the expense of maintaining accuracy and accountability suggest it will serve more as a gradual supplement than a full replacement in the short term.