Meta invests $14 billion in AI with Scale AI, but still lags behind

Meta has made a multibillion-dollar bet on artificial intelligence, spending more than $14 billion to hire Alexandr Wang, founder of Scale AI, and a group of his top engineers. The goal was to reposition the company in the AI race, which has so far been dominated by OpenAI, Anthropic, and Google. However, a year after this investment, the results are mixed: although Meta has managed to launch Muse Spark, its first proprietary foundation model, it is still far from the top and faces serious credibility challenges.

What happened?

According to CNBC, Wang's main achievement was the delivery of the Muse Spark model in April, marking Meta's first leap into proprietary foundation models, abandoning its strict adherence to open source. Previously, Meta had bet on the Llama family of models, offering an open-source approach that allowed developers to freely modify the models. However, the launch of Llama 4 in April last year was a failure, failing to captivate developers and leading Zuckerberg to reconsider his strategy.

Since then, Meta has introduced new subscription plans related to AI and business, as part of an effort to diversify its revenue beyond online advertising. But historically, this has not worked: Meta still relies on ads for 98% of its revenue.

Why is it important?

Meta's investment is one of the largest in the AI industry, but the market is not impressed. Meta's shares have fallen 18% in the last twelve months, the worst performance among mega-caps, along with Microsoft. This despite Meta reporting 33% revenue growth in the first quarter, the fastest rate since 2021.

Analyst Ralph Schackart of William Blair notes that investors are looking for tangible evidence of a growing list of AI-based products, even if monetization is slow. But for now, Meta lacks those concrete use cases.

Consequences and challenges

One of Meta's biggest hurdles is developer distrust. Rob May, CEO of Neurometric, says the AI community largely ignores Meta at this point. Krish Subramanian, CEO of KOI AI, adds that developers are more excited about Google's models than what Meta offers. Llama's appeal was precisely for developers seeking alternative open-weight models, but with Muse Spark, Meta has made little effort in that direction.

Subramanian warns that the lack of developer trust could hurt Meta if it does not focus on them, and notes that it took Microsoft years to regain the trust of open-source developers during the early days of Azure.

Additionally, Zuckerberg's metaverse ambitions have generated total losses of more than $80 billion since late 2020, making the AI bet a harder sell. Howard Yu, a business professor at the International Institute for Management Development in Switzerland, comments that Zuckerberg is running out of credibility margin and that his foray into virtual reality may have burned much of his goodwill among investors.

What should readers know?

Meta remains committed to the open-source ecosystem, according to a spokesperson, and plans to offer access to Muse Spark's underlying technology via an API, with testing already underway and a launch scheduled for this month. However, for the $14 billion investment to pay off, Meta needs to demonstrate that it can adopt and commercialize new AI products, beyond the positive impact AI already has on its advertising models.

In summary, Meta has taken an important step by moving away from open source and creating a proprietary model, but developer distrust and investor pressure could slow its progress. The company will have to work hard to regain lost credibility and show concrete results.