For Mark Zuckerberg, the artificial intelligence arms race has become a high-stakes battle, but recent reports suggest that the road to ‘super-intelligence’ is bumpier than expected. According to sources close to the company, Meta has decided to delay the release of its latest AI model, codenamed ‘Avocado’. Originally planned for the first quarter, the debut has been pushed back until at least May, shedding new light on the Menlo Park-based tech giant’s internal challenges.
The reason for this decision is not logistical, but pure performance. Current tests place the ‘Avocado’ in a perplexing position – the model offers capabilities greater than Google’s Gemini 2.5, but is still inferior to the upcoming Gemini 3. Being ‘in-between’ generations of competition is a risky place for a company that intends to spend a record $115-135 billion on infrastructure and proprietary chip development this year.
The delay is a signal that Meta is abandoning haste in favour of quality. The company does not want to repeat the mistakes of the past by releasing a tool that does not dominate its market rivals. However, this strategy comes at a cost. Every month of delay increases the gap with OpenAI and Google, and thus casts doubt on the rate of return on the huge capital investment. A Meta spokesperson tones down the mood, announcing that the upcoming model is primarily intended to show a ‘rapid trajectory’ of development, suggesting that the company is betting on systematic, frequent updates rather than a single, static breakthrough.
The most intriguing thread, however, remains the desperation to find bridging solutions. It has been reported that the leaders of Meta’s AI division were considering temporarily licensing Gemini technology from Google to power their own products. While final decisions have not been made, the mere fact that such talks are being held by a company seeking full technological independence shows the enormous pressure on Zuckerberg’s team. The industry will be keeping a close eye on the May launch window, which will show whether Meta is able to leap over the bar set by its competitors on its own, or whether the billions of dollars of investment will take even longer to yield the expected results.
