In a significant development, Meta Platforms has announced a delay in the release of its flagship AI model, Behemoth, which was initially set to debut in June. The rollout is now expected to occur in the fall or later, according to reports from The Wall Street Journal and other sources. This delay is attributed to concerns over the model’s performance, a situation that has sparked internal discussions about the direction of Meta’s substantial investment in AI.

Behemoth, a 400 billion parameter model, is seen as a critical component of Meta’s AI strategy, though it is smaller compared to OpenAI’s latest GPT models, which are rumored to be around a trillion parameters. Gene Munster, Managing Partner at Deepwater Asset Management, views the delay as a “black eye” for Meta but emphasizes that it does not alter their long-term opportunity in the AI space. Munster notes, “It speaks to how hard this is,” highlighting the challenges in developing advanced AI models.

Despite the setback, Munster points out that Meta has seen benefits from AI at scale, such as a 6% growth in daily active users in the March quarter, which is attributed to better content targeting enabled by AI. This growth indicates that Meta is making progress in leveraging AI, even if not at the pace initially anticipated.

The delay in Behemoth’s release underscores the complexity and difficulty in advancing AI technologies. Munster suggests that investors should temper their expectations about the rapid development of these models, as significant improvements are not happening as quickly as some had hoped. He also notes that companies like Nvidia are likely to benefit from the increased need for hardware to support AI development, as advancing these models often requires substantial computational resources.

Munster’s perspective on the delay is that while it presents a challenge, it is not a game-changer for Meta’s overall AI strategy. He believes that Meta, along with other major tech companies, will continue to invest heavily in AI, given the potential rewards. However, achieving these rewards will require patience, as the development of high-performance AI models is a complex and time-consuming process.

The delay also highlights the competitive landscape in AI, where companies like OpenAI are pushing the boundaries of model size and capability. Meta’s decision to delay Behemoth reflects the high standards set by the public and investors for AI model performance. The company’s focus on ensuring that Behemoth meets these expectations suggests that Meta is committed to delivering quality rather than rushing to meet earlier timelines.

In conclusion, while the delay of Behemoth presents a short-term setback for Meta, it does not diminish the company’s potential in the AI sector. The challenge of developing advanced AI models is significant, but Meta’s ongoing investment and strategic focus on AI suggest that the company remains committed to achieving its long-term goals in this area.