.png)
A historic day for artificial intelligence: Monday, December 2, 2025 will remain engraved as a major turning point in the war of technological giants. As Apple announces the departure of its AI manager and appoints a former Google and Microsoft executive to catch up, OpenAI is raising maximum alert in the face of Gemini's spectacular rise in power. Two simultaneous events that reveal the unprecedented ferocity of competition in the generative AI sector.
Apple has officially announced the departure of John Giannandrea, vice president in charge of artificial intelligence and machine learning since 2018, who will remain an advisor until he retires in early 2026. In its place, the apple brand names Amar Subramanya, a particularly noted profile in the industry.
Amar Subramanya spent 16 years at Google where he participated directly in the development of Gemini and Imagen 3, two key projects in Google's strong comeback against OpenAI. More recently, he held the position of corporate vice president of AI at Microsoft during the year 2025, in the midst of a period of massive integration of artificial intelligence into Windows and Copilot.
Tim Cook praised the role John Giannandrea played in building and advancing Apple's work in AI, while saying that Amar Subramanya would bring his “extraordinary expertise” in the field.
The change is taking place in a particularly tense context. Apple Intelligence, available in French since April on recent iPhones starting with 15 Pro, allows you to edit photos, rewrite or summarize texts and generate answers. But the deployment was not without a hitch.
The tool responsible for summarizing the notifications generated several erroneous claims, sometimes erroneously relayed as proven facts, leading to public criticism from organizations such as the BBC. These embarrassing mistakes created a climate of distrust in a technology that was supposed to represent the future of the Apple ecosystem.
According to several internal sources, the problems are deeper than they seem. An external investigation highlighted internal structural tensions with limited coordination between AI and marketing teams, differences in budgetary goals, and fragile leadership. Some employees have even ironically nicknamed the department “AI/Mless,” a sign of a widespread loss of trust.
The new AI-boosted Siri, promised for next spring, would still not be one hundred percent up to date. At the beginning of the year, Apple delayed the release of the improved version of its Siri voice assistant, promising it by 2026. The challenge is immense for newcomers who must not only correct current mistakes but also bridge a technological gap estimated at several years.
A wave of departures to competing companies such as OpenAI, Meta or Google has weakened Apple's technical capabilities. In this context, an important part of Giannandrea's teams is joining the fold of Eddy Cue, head of Services including Siri, Spotlight, Apple Music and the entire Apple Intelligence ecosystem.
Apple's preferred model is based on the local processing of AI tasks using Apple Silicon chips to maintain user privacy, with Private Cloud Compute that deletes data immediately after use. However, this differentiating approach raises a central question: can it compete with competitors who operate huge data centers to train much larger models?
In an ironic twist of events, OpenAI is now in the position that Google was in three years ago. CEO Sam Altman said in a memo to employees that the company needs to improve the chatbot and is delaying other products in the meantime.
Sam Altman considers that the company is going through a critical moment in the face of competition from Google Gemini 3, which surpasses GPT-5.1 on several benchmarks. This general mobilization follows the impressive performance of Google's latest model, launched last month.
Google's launch of its improved Gemini model last month, which surpassed OpenAI on industry benchmark tests and led to a surge in the search giant's share price, was a key factor in OpenAI's increased sense of urgency.
To focus all resources on improving ChatGPT, OpenAI is taking drastic measures. OpenAI has suspended the deployment of ads in ChatGPT, as well as several projects like buying agents and the Pulse morning assistant.
Altman encouraged temporary team transfers and added that OpenAI would be holding a daily call for employees responsible for improving ChatGPT. Nick Turley, head of ChatGPT, confirmed on X that the aim is to make the chatbot “even more intuitive and personal.”
The three priorities identified are clear: improving the speed and reliability of ChatGPT, allowing it to answer a wider range of questions, and strengthening personalization options for users.
The numbers speak for themselves. Gemini's user base grew from 450 million in July to 650 million in October, a growth that OpenAI is clearly worried about. In comparison, ChatGPT has stagnated at 800 million weekly users since the end of summer, far from the target of one billion by the end of 2025.
Gemini 3 set a new record on Humanity's Last Exam, a test created by AI security researchers to identify artificial superintelligence. The model has also been widely praised for its Nano Banana image generator and deep reasoning abilities.
Beyond pure technology, OpenAI faces a major handicap. OpenAI's financial situation presents a challenge, as the company depends on constant fundraising to support its operations, unlike its main competitors like Google who can fund their investments thanks to their massive revenues.
OpenAI's aggressive spending means the company will need to generate around $200 billion in revenue to reach profitability by 2030, according to its own financial projections. In November, OpenAI announced that it was on track to reach over $20 billion in annualized revenue this year.
The competition isn't just coming from Google. Anthropic has shown strong momentum among businesses, with more than 300,000 business customers in September, compared to less than 1,000 just two years ago. Large accounts, defined as generating over $100,000 in annual revenue, have grown sevenfold over the past year.
The current situation of OpenAI is ironically reminiscent of that of Google in early 2023, when the launch of ChatGPT prompted Sundar Pichai to trigger his own red alert. Today, the roles are reversed: Google controls the entire technology chain, from proprietary TPU chips to integration into its 3 billion Android devices, while OpenAI depends on Microsoft's cloud infrastructure and expensive Nvidia processors.
These two simultaneous events illustrate the extraordinary intensity of the war for talent in artificial intelligence. The recruitment of Amar Subramanya by Apple, just a few months after joining Microsoft, testifies to this unprecedented volatility. The best AI researchers now command salaries well in excess of $10 million annually.
Altman's use of a system of color codes: yellow, orange and red to prioritize the urgency of issues shows how critical the situation has become. The “code red” had never been triggered before at OpenAI, underlining the magnitude of the perceived threat.
These upheavals of December 2, 2025 raise fundamental questions for the future of artificial intelligence:
For Apple : Can the appointment of Amar Subramanya turn the tide and allow Siri to catch up with ChatGPT and Gemini? Will the privacy-based approach be a differentiating advantage or a disadvantage in the race for performance?
For OpenAI : will the “code red” allow us to regain the advantage over Gemini? Can the company maintain its leadership without the profitability and structural resources of Google? A new AI model more powerful than GPT-5.1 would be expected as early as next week to counter Google's advance.
For industry : will this frantic race result in applications that are really useful for end users, or are we witnessing an ego war between technological giants? Is the concentration of talents and resources between a few actors healthy for innovation?
The spring of 2026 promises to be decisive: it will be the time for the new Apple Siri, potentially new OpenAI models, and undoubtedly new advances from Google. The battle is far from over, and users around the world are watching closely which giant will be able to turn the promise of AI into a concrete and reliable reality.
