OpenAI plans to spend $1t despite having $13b annual revenue

OpenAI is generating about US$13 billion in annual revenue, with 70% coming from consumer subscriptions to ChatGPT, according to the Financial Times.

The company’s user base includes 800 million regular users, but only 5% are paying subscribers.

OpenAI has committed to spending over US$1 trillion in the next decade, and has recently made deals for more than 26 gigawatts of computing capacity with Oracle, Nvidia, AMD, and Broadcom.

To address the gap between revenue and planned spending, OpenAI is reportedly considering new business areas such as government contracts, shopping tools, video services, consumer hardware, and supplying computing power through its Stargate data center project.

Several major US companies rely on OpenAI for large contracts, which could affect broader market stability if OpenAI faces financial challenges.

🔗 Source: TechCrunch

🧠 Food for thought

Implications, context, and why it matters.

OpenAI’s compute commitments dwarf its revenue, raising questions about contract structures

OpenAI makes $13 billion a year 1 while infrastructure promises top $1 trillion 2, yet filings do not clarify take-or-pay terms, revenue triggers, or financing contingencies 3. The list spans a $300 billion Oracle pact over five years starting in 2027 4, about $22.4 billion with CoreWeave (a specialized cloud provider focused on AI workloads) 5, and roughly $10 billion with Broadcom for custom AI chips 3.
Management projects about $44 billion in cumulative losses before 2029 profitability 6 and plans roughly $60 billion a year for Oracle compute (cloud data center capacity for training plus running AI models) 3.
Oracle logged more than $317 billion in future contract revenue in Q1 4, while reports put OpenAI on a path to a $500 billion valuation 6.

Enterprises dependent on OpenAI face concentration risk, creating demand for multi-model resilience tools

Large US firms lean on OpenAI for core work 1 and about 20 million pro subscribers use paid services 7, so a disruption tied to finances could leave teams without backups. That gap fuels demand for multi-LLM routing (software that distributes AI requests across providers), failover systems (automatic switching to backup services), and cost optimization.
Product leaders can find ChatGPT Enterprise users through LinkedIn postings that mention OpenAI integrations. Platform teams can pitch tools that spread workloads across Anthropic (the AI company behind the Claude assistant), Google Gemini (Google’s flagship AI model), and open-source models. Investors can back LLM orchestration startups that track uptime (service availability), switch providers during outages, and lower costs as buyers weigh single-vendor risk tied to $44 billion projected losses before 2029 6.