Open AI – Code Red

The technology battle is over. The battle for the user begins

• Three years later, and the shoe is on the other foot as now it is OpenAI’s turn to issue a code red, making it Google’s turn to make a rival dance.
• This also fires the starting gun on the race to become the AI ecosystem provider, and although OpenAI remains a long way out front, Google is snapping at its heels with plenty of others not far behind.
• Sam Altman has declared a “code red” situation at OpenAI and directed that resources be focused on ChatGPT away from other projects to ensure that it continues to be the go-to place to access AI services.
• This is a result of recent wins by Google, where Gemini 3 outperforms GPT-5 on the usual benchmarks, users have grown to 650 MaU and app downloads of Gemini are now at 73m per month compared to ChatGPT at 93m.
• All sorts of reasons can be trotted out to explain these figures, such as gaming of benchmarks, and most users have ChatGPT already, but the reality is that OpenAI’s technical lead has been greatly diminished, if not altogether gone.
• This means that arguing over who has the best model on one benchmark or another is pointless, as it will now be down to the users to decide which service they wish to use, where they spend the time and to whom they give their money.
• This is why Altman is directing resources to focus on things like the user experience, personalisation, image generation, and so on, which are features aimed at delighting and engaging users, not on advancing the cause of super-intelligent machines.
• This is what RFM Research has long called AI Ecosystem 2.0 (see here) and it has very recently published research assessing how OpenAI is addressing this issue and what else it needs to do (see here) to win the battle to become the go-to AI Ecosystem.
• This is a battle that OpenAI must win or face going out of business because, as it stands today, even before spending hundreds of billions on multi-gigawatt data centres, it is already burning billions of dollars every quarter to offer its services.
• This is because the vast majority of its 800m WaUs do not pay it anything at all, all the while the compute to provide the “free” service is increasingly expensive.
• This is where Google has two key advantages.
• First, its monetisation method is well established and already generates plenty of cash and second, Google searches executed without an LLM cost a fraction of what a search on OpenAI costs.
• This is why AI mode and AI Overviews are largely driven by searches that are then stitched together by an LLM, with the more expensive LLM system being employed only when more details are required.
• This gives Google a great economic advantage over OpenAI, but at the moment, OpenAI has the overall edge.
• This is because ChatGPT is a global household name, meaning that anyone who wants to ask AI a question will always think of ChatGPT first.
• This is reflected in the user number, which, when adjusted into the MaUs that Google and everyone else refer to, is probably somewhere around 1.2bn.
• This means that developers who are creating AI services will prioritise GPT-5 over Gemini due to the larger addressable market that they will reach.
• However, with OpenAI’s developer tools only just having launched, its program remains fairly nascent, and it now needs to rapidly gain traction among developers if it is to keep attracting users and keep them out of the hands of Google.
• This is a must-win for OpenAI as a thriving digital ecosystem has been shown to deliver a multi-trillion dollar valuation, which is precisely what OpenAI needs if it is going to keep its investors happy.
• Failure will mean it will become unable to pay its bills, build its infrastructure, and, crucially, it will become difficult and expensive to raise more money.
• With the lead that OpenAI has established in the market, it is still the most likely to end up the overall winner, but this is by no means guaranteed.
• Google, Anthropic, Perplexity, Microsoft, Amazon, Alibaba, DeepSeek, and many others are all trying to make a splash in this market, meaning the competition will be intense.
• From a risk-reward perspective, paying $500bn for OpenAI looks dangerous when one can have Google, which is the clear No.2, is already economically viable and trades at a fraction of the price.