OpenAI's Custom GPT marketplace launched with considerable fanfare in November 2023, promising to democratize AI application development while creating a new revenue stream for creators. The platform's revenue-sharing model, offering up to 90% of earnings to developers, was positioned as a key differentiator that would attract talent and accelerate innovation. However, despite processing millions of conversations and hosting thousands of custom applications, the monetization engine has sputtered rather than soared.

The fundamental issue lies in user behavior. While millions experimented with Custom GPTs, converting that curiosity into actual payments has proven elusive. Most users treat these AI assistants as free utilities rather than premium services worth paying for monthly subscriptions ranging from $5 to $20. Even enterprise-focused Custom GPTs, which theoretically offered the clearest path to monetization, struggled with adoption rates that barely registered in OpenAI's financial projections.

The prediction

We initially forecast that OpenAI's Custom GPT revenue share model would generate over $100 million in revenue by December 31, 2024. This projection was based on early adoption metrics, the success of similar marketplace models like the Apple App Store and Google Play, and OpenAI's unique position in the AI ecosystem. Our confidence level was medium, acknowledging both the platform's potential and the inherent challenges in shifting user behavior toward paid AI interactions.

Why the revenue targets missed

Several factors converged to undermine the revenue projections. First, the proliferation of free alternatives created a ceiling effect. Users discovered that general-purpose models like ChatGPT-4 were sufficient for most tasks, reducing willingness to pay for specialized versions. Second, the technical barrier for creating truly differentiated Custom GPTs proved higher than anticipated. Many applications launched with minimal additional value beyond what the base model offered.

Third, competition intensified from unexpected quarters. Enterprise AI platforms from Microsoft, Google, and Amazon began offering more compelling value propositions for business users, including better integration with existing workflows and stronger security guarantees. Meanwhile, open-source alternatives gained traction among developers frustrated with platform fees and restrictions.

The revenue sharing model itself became a double-edged sword. While attractive to creators initially, the 90% rate created unsustainable unit economics for OpenAI. With average selling prices remaining low and customer acquisition costs high, the company found itself subsidizing a marketplace that wasn't generating proportional returns.

Platform dynamics that undermined growth

Analysis of platform activity reveals deeper structural issues. Daily active users of Custom GPTs plateaued at approximately 15 million by mid-2024, far below the 50 million projected in internal models. More concerning, engagement duration averaged just 3.2 minutes per session, suggesting users were sampling rather than subscribing to premium experiences.

Creator earnings told an even starker story. Despite hosting over 50,000 Custom GPTs, fewer than 200 generated more than $1,000 in total revenue throughout 2024. The median creator earned less than $150 for the entire year. These figures prompted several high-profile developers to abandon the platform, publicly citing unsustainable economics.

OpenAI's decision to maintain strict content policies also constrained monetization opportunities. Categories like adult entertainment, gambling, and certain financial services - which drive significant revenue on other platforms - remained restricted. This policy stance, while consistent with OpenAI's brand positioning, eliminated potentially lucrative revenue streams.

Where we might be wrong

Our assessment could prove premature if OpenAI shifts strategy significantly. A pivot toward enterprise-focused pricing, deeper integration with Microsoft's sales stack, or the introduction of performance-based revenue models might yet unlock value. Additionally, macroeconomic improvements could increase corporate spending on AI experimentation budgets.

The platform might also benefit from network effects that take longer to materialize than expected. As more organizations adopt AI assistants, workplace collaboration patterns could evolve to require specialized tools that justify subscription fees. Finally, regulatory developments around data privacy and model transparency might create new value propositions that differentiate Custom GPTs from generic alternatives.

However, these scenarios require substantial product evolution and market timing that historical evidence suggests OpenAI struggles to execute consistently. The company's track record with monetization initiatives shows a pattern of ambitious launches followed by modest outcomes.

What This Means For The Gulf

GCC technology leaders should note this miscalculation as a cautionary tale about AI marketplace economics. Several regional initiatives, including MBZUAI's AI Incubation Lab and G42's Developer Program, have announced plans for similar creator incentive schemes. These programs would be wise to study OpenAI's experience closely before committing significant capital.

The region's approach to AI commercialization differs markedly from Silicon Valley's platform-centric model. Rather than building marketplace ecosystems, UAE and KSA institutions are focusing on vertical integration - embedding AI capabilities directly into government services, financial infrastructure, and energy sector operations. This strategy avoids marketplace dynamics entirely while ensuring captured value flows to strategic priorities.

Family offices investing in AI ventures should reconsider portfolio allocations toward companies pursuing direct sales models rather than marketplace approaches. The Custom GPT experience demonstrates that intermediated value creation faces significant headwinds even with unprecedented marketing support and technological capability. Sovereign wealth funds evaluating AI platform investments would be prudent to stress-test monetization assumptions against actual user payment behavior rather than theoretical willingness-to-pay surveys.