In the world of technology, having size, money, and resources does not always suffice to win all battles. Meta is proving this in the race for artificial intelligence. Despite having put at least $14.3 billion on the table, the company continues to struggle to position itself as a leading reference in chatbots and advanced models.
The most recent and striking move was the investment in Scale AI, a company specialized in data to train and improve AI systems. Meta acquired almost half of the company and, as a bonus, added its founder, Alexandr Wang, to lead efforts in superintelligence.
This bet was not only for labeled and curated data but also for key talent to accelerate internal development. It came at a complicated time, following the lukewarm reception of Llama 4 and with competitors like Google, OpenAI, and others setting the pace.
The new model and its deployment
A visible result of this stage is Muse Spark, the first model from the new family of Meta Superintelligence Labs. The company has already integrated it into Meta AI and is bringing it to WhatsApp, Instagram, Facebook, Messenger, and even its augmented reality glasses.
The great advantage of Meta is its enormous user base and the ability to distribute innovations directly on platforms that people already use every day. It does not need users to download a new app. However, turning that presence into habitual use and relevance is another story.
Muse Spark shows good results in language and visual comprehension, but falls behind in coding and abstract reasoning compared to market leaders. For now, it does not seem to be changing user habits when they think of a powerful chatbot.
