Google has OpenAI rattled. Why that’s a problem for Oracle and Microsoft.

Google released its Gemini 3 AI model last month and it made a splash, receiving rave reviews from users and analysts, and beating OpenAI’s latest GPT 5.1 in benchmark tests.

OpenAI doesn’t just accept this. The startup has declared a “code red” effort to develop ChatGPT and will delay other initiatives, such as AI agents for advertising, healthcare and shopping, and a personal assistant, to redirect resources, according to an internal memo from OpenAI CEO Sam Altman on Tuesday as reported by The Wall Street Journal.

OpenAI did not immediately respond to a request for comment early Tuesday. Barron’s owner News Corp has a content licensing partnership with OpenAI.

The concern for Oracle and Microsoft is that OpenAI’s efforts to keep pace with Google will push back its profit-making timeline and raise questions about its sustainability.

OpenAI has made massive spending commitments totaling more than $1 trillion. These include a $300 billion cloud computing deal with Oracle and a $250 billion deal with Microsoft. OpenAI is valued at approximately $500 billion, and annual revenue is expected to exceed $20 billion this year.

OpenAI’s backers are confident that the company will be able to generate significantly more revenue from advertising, subscriptions and other sources to fund its commitments. Altman suggested that OpenAI could reach $100 billion in annual revenue by 2027. But that trend could be derailed if Google’s Gemini starts to take significant market share and OpenAI is forced to delay its monetization efforts.

Gemini’s monthly active users increased from 450 million in July to 650 million. This compares with ChatGPT’s 800 million weekly active users, down from nearly 700 million in July. According to Bloomberg, this indicates that Google’s AI is growing faster, and Gemini is also ready to power Apple’s digital assistant Siri starting early next year.

While it’s difficult to compare AI models across different domains, the Gemini 3 Pro mode appears to outperform OpenAI’s GPT 5.1 on general intelligence and coding-specific tasks, according to data from AI benchmarking firm Artificial Analysis, as well as xAI, Meta Platforms, and rivals from China’s Alibaba and DeepSeek, but it’s also more expensive to use.

OpenAI isn’t just facing pressure from Google. Chinese AI start-up DeepSeek launched two new large language models on Monday. It said DeepSeek-V3.2 achieved similar performance to OpenAI’s GPT-5 across multiple benchmarks, while a compute-intensive variant surpassed GPT-5 and matched the Gemini 3 Pro in terms of reasoning capabilities.

“There’s a lot to be excited about in the models and technical documentation, raising our expectation that DeepSeek-4 will be released in the near future,” DA Davidson analyst Alexander Platt wrote in a research note. he wrote.

A tough task awaits OpenAI to keep it ahead of the pursuing pack. Microsoft and Oracle will be watching closely.

Write to Adam Clark at adam.clark@barrons.com