Meta’s Avocado AI Delayed as Company Reportedly Considers Licensing Go

When Meta hired Alexandr Wang nine months ago and poured $14.3 billion into building an AI super team, the message was clear: the company would spare no expense to close the gap with OpenAI and Google in the artificial intelligence race. This week, that ambitious plan encountered a significant setback.

Meta’s latest AI model, code-named Avocado, has been delayed until May—missing its planned March launch—according to a report in the New York Times. While the model reportedly outperforms Meta’s previous efforts and even Google’s Gemini 2.5, it falls short of Gemini 3.0, which launched in November, as well as the latest models from OpenAI and Anthropic.

“The notion of Mark Zuckerberg asking Google’s Sundar Pichai for permission to use Gemini is almost impossible to imagine.” — Industry Analyst

The Unthinkable Partnership

Perhaps more surprising than the delay itself is what Meta’s leadership has reportedly discussed internally: licensing Gemini from Google on a temporary basis until Avocado is competitive. The two companies are not casual competitors—they are arch-rivals battling across advertising, online video, smart glasses, and now artificial intelligence.

The discussions, first reported by the New York Times citing anonymous sources, highlight the depth of Meta’s AI challenges. It’s unclear whether Meta executives have actually approached Google or merely discussed the possibility internally. Either way, the fact that such a conversation is happening represents a significant strategic pivot.

This wouldn’t be Google’s first time providing AI infrastructure to a major rival. Apple announced earlier this year that it would use Gemini to power Siri, a move that surprised many industry observers. But Meta and Google have historically maintained a more adversarial relationship, making any potential partnership particularly notable.

The $14.3 Billion Question

The talent investment has been substantial. Alexandr Wang, the 27-year-old founder of Scale AI, was hired to lead Meta’s AI efforts alongside a team of top researchers and engineers. The $14.3 billion allocated to this initiative represents one of the largest AI investments in corporate history.

The technical gap persists despite these resources. While Avocado shows improvement over Meta’s previous models, the AI landscape has shifted rapidly. Google’s Gemini 3.0, OpenAI’s GPT-5 series, and Anthropic’s Claude models have all raised the bar for what constitutes state-of-the-art performance.

The strategic implications extend beyond Meta’s internal products. The company’s AI models power everything from content recommendations to its emerging metaverse applications. Falling behind in foundational model capabilities could cascade through Meta’s entire product ecosystem.

“We’re building the infrastructure to power the next generation of AI experiences. This isn’t just about cost savings—it’s about having the flexibility to innovate at the pace the technology demands.” — Meta Engineering Blog

What This Means for the AI Landscape

Meta’s struggles illustrate a broader challenge facing the AI industry: building competitive large language models is extraordinarily difficult, even with virtually unlimited resources. The technical expertise, training infrastructure, and research insights required to compete at the frontier are concentrated among a handful of organizations.

For Google, a potential licensing deal with Meta would represent both validation and strategic positioning. Gemini’s adoption by two of its biggest rivals—Apple and potentially Meta—would establish it as the de facto standard for companies that can’t or won’t build their own foundational models.

The situation also raises questions about Meta’s long-term AI strategy. If the company licenses Gemini while continuing to develop Avocado, it risks creating internal confusion about priorities and resource allocation. If it abandons Avocado entirely, it surrenders control of a critical technology layer to a direct competitor.

The Road Ahead

Industry observers are watching closely to see how Meta navigates this challenge. The company has historically been willing to make bold bets and endure short-term setbacks for long-term positioning. But the AI race moves faster than previous technology transitions, and delays can compound quickly.

The next two months will be critical. If Avocado can close the performance gap by its revised May timeline, Meta may be able to maintain its independent AI strategy. If not, the company may face a difficult choice between licensing from Google or accepting a permanent position behind the AI frontier.

For now, one thing is clear: the AI race has entered a new phase where even the tech giants with the deepest pockets are struggling to keep pace. Meta’s $14.3 billion bet is far from settled.

This article was reported by the ArtificialDaily editorial team. For more information, visit Fortune.