Meta Consolidates AI Efforts Under TBD Lab
Meta Platforms has strategically formed
TBD Lab, a research unit comprised of a few dozen top researchers and engineers, to accelerate its ambitions in artificial intelligence. This team, highlighted by CFO Susan Li at the Goldman Sachs Communacopia + Technology conference, is tasked with developing Meta's next generation of foundation models over the upcoming one to two years[2]. The name “TBD Lab,” initially a temporary placeholder, was retained to reflect the evolving nature of its mission and output[2].
- TBD Lab is a core part of Meta AI assistant development and broader AI innovation strategy[2][3].
- The unit is described as “very talent-dense,” combining elite researchers and engineers in a small, agile team[2][1].
A New Structure for AI Superintelligence
In response to recent challenges, such as middling reception for the
Llama 4 open-source model, Meta has reorganized its AI operations under the umbrella of
Meta Superintelligence Labs (MSL). The new division includes four groups:
- TBD Lab, now spearheading next-generation foundational model research
- A products team, which oversees the Meta AI assistant
- An infrastructure team focusing on scaling AI capabilities
- The Fundamental AI Research (FAIR) lab, dedicated to long-term innovation[2][4]
TBD Lab's Research Focus
TBD Lab’s research and engineering efforts will be closely coupled with the
FAIR lab, with FAIR serving as an “innovation engine” by funneling fundamental advances directly into the training of new models within TBD Lab[3]. The lab is reportedly exploring novel AI architectures, potentially including an "omni model" capable of multimodal processing (text, visual, audio, and more)[3].
Aggressive Talent Acquisition and Investment
Meta’s CEO, Mark Zuckerberg, is closely involved in talent strategy, utilizing direct outreach and lucrative compensation packages (reportedly up to $1 billion in signing bonuses for key recruits) to attract leading minds in artificial intelligence away from top competitors[2][4]. Notably, Meta’s acquisition of talent includes high-profile individuals, such as former Scale AI CEO Alexandr Wang, following a $15 billion purchase of Scale AI[4].
Next Steps: Llama Model Development and Beyond
One immediate objective for the TBD Lab is to lead development of the next major release of the
Llama large language model series, positioning Meta more competitively alongside AI giants like
ChatGPT,
Anthropic,
Google Gemini, and
DeepSeek[5][4]. The restructuring also dissolves the older AGI Foundations team, dispersing its members to the new units within MSL but not directly to TBD Lab itself[3].
Meta’s Future in AI
With TBD Lab at its core, Meta’s AI ambitions now include broad, aggressive investments in research talent, infrastructure, and next-generation AI models. The company's reorganized approach reflects both lessons learned from previous setbacks and a belief that a focused, elite team can accelerate progress towards superintelligence—a key component in Zuckerberg’s broader vision where AI enhances personal fulfillment, not just enterprise productivity[4].
