Meta Platforms has achieved a significant milestone in its artificial intelligence ambitions, with the company’s Superintelligence Labs delivering its first AI models for internal use, according to Chief Technology Officer Andrew Bosworth.
Speaking with Reuters at the World Economic Forum in Davos, Switzerland, on Wednesday, Bosworth said the models showed promise despite the team’s relatively short development timeline.
“They’re basically six months into the work, not quite even,” Bosworth said, describing the models as “very good.”
New Research Division Shows Early Results
The delivery of these first models represents tangible progress from Meta’s dedicated AI research division, which was established as part of a broader company restructuring aimed at competing with rivals like Google, OpenAI and Microsoft in the generative AI space.
While Bosworth did not specify which models were delivered, reports from December indicated Meta was developing multiple AI systems. These include a text-focused model codenamed Avocado, scheduled for a first-quarter debut, and an image and video generation model called Mango.
The announcement comes at a critical time for Meta’s AI strategy. The company has faced criticism over the performance of its Llama 4 model compared to competing systems from Google and other tech giants. Bosworth acknowledged the ongoing challenges, telling Reuters that significant work remains on the technology.
“There’s a tremendous amount of work to do post-training to actually deliver the model in a way that’s usable internally and by consumers,” he said.
Aggressive Talent Acquisition and Leadership Changes
Meta’s AI push has involved substantial organizational changes over the past year. The company overhauled its AI leadership structure, launched Superintelligence Labs, and embarked on an aggressive recruitment campaign that included multi-million dollar compensation packages to attract top researchers and engineers.
These investments reflect CEO Mark Zuckerberg’s determination to position Meta as a leader in AI development, even as the company’s approach differs significantly from its competitors.
Different Path to AI Monetization
Unlike rivals Amazon, Google and Microsoft, which have established clear revenue streams from their AI investments, Meta has focused primarily on internal applications. The company uses AI to power engagement algorithms, recommendation engines, ad targeting systems and features like Meta AI and Reels.
Microsoft monetizes OpenAI’s models through Azure cloud subscriptions and Copilot services. Google sells access to Gemini and Vertex AI through its cloud division while building a growing tensor processing unit business. Amazon generates recurring enterprise revenue through its Bedrock and SageMaker platforms.
Meta’s strategy centers on improving user metrics and enhancing existing products rather than creating standalone AI services. Whether this approach will generate comparable returns remains an open question for investors and industry analysts.
Reality Labs Restructuring
The AI focus comes as Meta makes difficult decisions elsewhere. The company plans to cut 10% of positions within its 15,000-person Reality Labs division, which handles virtual reality products. The move signals a strategic shift away from VR toward AI-powered wearables and other emerging technologies.
This restructuring reflects Meta’s evolving priorities as it allocates resources toward AI development and deployment across its product ecosystem.
Looking Ahead
Bosworth’s comments at Davos suggest Meta remains committed to its AI strategy despite the challenges. The successful delivery of initial models from Superintelligence Labs provides validation for the company’s investment in building proprietary AI systems rather than relying exclusively on external technologies.
However, the path forward includes significant technical hurdles. Post-training work, safety considerations, alignment with user needs and competitive pressures all present ongoing challenges for Meta’s AI teams.
As the AI race intensifies among tech giants, Meta’s ability to translate its internal AI development into tangible product improvements and business value will determine whether its substantial investments pay off. The company’s extensive user base across Facebook, Instagram, WhatsApp and other platforms provides a potential advantage for scaling AI applications, but execution remains critical.
For now, the delivery of these first models marks progress in Meta’s transformation from social media platform to AI-driven technology company, a journey that continues to unfold as the competitive landscape evolves.
Stay informed on the latest AI developments, tech industry news, and emerging trends at NextGen Bulletin.
