Key Points:
- Meta pushed the release of its “Avocado” AI model from this month to at least May.
- Internal testing showed the new model currently underperforms compared to top competitor offerings.
- The company plans to spend up to 135 billion dollars on AI infrastructure this year.
- Meta leadership is considering a temporary deal to license Google’s Gemini AI.
Meta decided to delay the release of its highly anticipated artificial intelligence model, code-named “Avocado.” The company originally scheduled a launch for this month, but sources familiar with the matter say the product will not arrive until at least May. This delay marks a significant hurdle for the social media giant as it pours massive resources into the race for “superintelligence,” a future state where machines theoretically outsmart human beings.
The decision to hold back the software comes down to performance metrics. According to recent reports, the new model failed to meet internal benchmarks when compared to the latest systems released by competing AI labs. Meta has spent months developing this text-based model, which industry watchers previously expected to see in the first quarter of this year.
Meta spokespeople defended the company’s progress despite the setback. In an emailed statement, a company representative acknowledged the need for more polish. The company believes the next model will prove good, but they emphasized that the most important factor remains the rapid speed of their development. They promised to steadily push the frontier of their AI capabilities throughout the year as they release a succession of newer, better models.
To bridge the gap while they refine their own technology, Meta’s internal leadership teams have held discussions about a temporary partnership. Reports suggest that Meta might license Google’s Gemini AI to power its own products in the interim. While these talks are reportedly taking place, the company has not yet reached any final decision on whether to partner with a direct competitor.
The stakes for Meta are incredibly high. In January, the company announced massive capital spending plans ranging between 115 billion and 135 billion dollars for the year. This money funds everything from sophisticated AI training pipelines to a new roadmap for designing custom in-house computer chips. Meta clearly intends to remain a top player in the industry, even if individual model launches hit bumps in the road.
The push toward superintelligence requires consistent success, and the company is under immense pressure to prove its billions in investment are actually working. Investors remain focused on whether Meta can close the gap with OpenAI, Google, and other leaders in the field.
Market expectations for these companies are at an all-time high. Every delay or performance miss faces heavy scrutiny from Wall Street, which watches the sector’s capital expenditures and model quality with extreme caution. For now, Meta is asking for patience, signaling that their internal team continues to cook up new improvements that they expect to show off very soon.
Despite the recent rumors about internal struggles, the company insists it is on a rapid trajectory. The roadmap for 2026 remains aggressive, and the leadership team views these early releases as just the beginning of a long year of iterative improvements. Whether or not they choose to bring in outside technology like Google’s Gemini to help maintain that momentum will be a key storyline to watch over the coming months.
