Meta’s Strategic Brilliance: Llama 2 May Be Their New Social Graph

In a move that has caught the tech industry’s attention, Meta recently announced the release of Llama 2, the second version of its free, open-source large language model (LLM). I was delighted with this development as a user and developer of products utilizing large language models. However, delving deeper into Meta’s strategy, I was truly impressed by the company’s tactical and strategic brilliance. In this article, we will explore Meta’s decision to open-source Llama 2, its implications in the highly competitive landscape of ML-driven products, and how it addresses Meta’s unique position in the tech industry. 

The Battle of LLMs- Meta, Microsoft, and Google: Today, industry giants such as Microsoft/OpenAI and Google are competing fiercely to develop and market the best ML-driven products. This battle extends to the quality of their respective LLMs, with offerings like ChatGPT from OpenAI and Bard from Google. These companies fiercely guard their proprietary LLMs, treating them as prized assets. However, Meta’s decision to go open-source with Llama 2 represents a game-changing strategic move.

Tactical Advantage of Open-Source LLMs: By open-sourcing Llama 2, Meta gains a tactical advantage by preempting potential customers who might have turned to competitors like Google or Microsoft. Meta recognizes the value of attracting users and developers to its platform, leveraging the collaborative power of an open-source community. 

Additionally, opening up their LLM to others allows for valuable feedback, testing, and iterations that are crucial to refining and enhancing the model’s capabilities. While relatively harmless for smaller companies, these mistakes can significantly damage larger, publicly scrutinized entities. 

Strategic Implications for Meta: Meta, with its comparatively less diversified revenue stream among the major tech players, is seen by some analysts as more exposed or fragile. Unlike Apple or Microsoft, Meta does not sell software or hardware directly. Furthermore, it lacks the extensive commercial cloud infrastructure of Google, Amazon, or Microsoft. However, Meta’s investments in the Metaverse over the years have been a means to diversify beyond being a mere aggregator of advertising. The Metaverse holds potential for complex B2B and B2C offers, making Meta’s strategic bet on this frontier a nuanced and calculated move. 

The First Intangible Asset of Industrial Size: What truly sets Meta’s open-source LLM initiative apart is the strategic significance of LLMs themselves. Building a large language model requires an enormous training set, teams of exceptional research scientists, and costly hardware to facilitate training and prediction. Consequently, LLMs can be considered the first intangible asset of industrial size. Microsoft’s $10 billion investment in OpenAI is a testament to the substantial resources required to develop LLMs. As such, only a select few companies can build such models. Furthermore, monetizing LLMs can be achieved through various business models post-free release. 

The Asset that Drives Diversification: By harnessing the expertise of Yann Lecun’s exceptional research team, Meta has acquired a formidable asset in Llama 2. This LLM allows Meta to diversify beyond its primary revenue source, advertising. By expanding its offerings to include innovative solutions based on Llama 2’s capabilities, Meta can tap into new markets and revenue streams. This strategic move not only bolsters Meta’s position in the industry but also affords the company the potential to compete more effectively with its tech counterparts. Meta’s decision to release Llama 2 as an open-source LLM demonstrates tactical and strategic brilliance. It enables Meta to gain a competitive edge by preempting potential customers and benefiting from community-driven development. Moreover, by investing in Llama 2 and capitalizing on its vast potential, Meta positions itself to diversify beyond its traditional revenue streams. As the tech landscape evolves, Meta’s open-source LLM initiative represents a forward-thinking approach that showcases the company’s commitment to innovation, collaboration, and long-term success in an ever-changing industry.

VP - AI Science & Research at Quincus

Christophe Pennetier is the Chief AI & Optimization Scientist at Quincus. As the Chief AI & Optimization Scientist, he brings 15+ years of global experience as a technical lead and a passionate polymath with expertise in bridging research scientists and application/innovation teams. Christophe's expertise in AI, NLP, and data science is complemented by a talent for strategy, creativity, and execution

[Announcing Gretel Navigator] Create, edit, and augment tabular data with the first compound AI system trusted by EY, Databricks, Google, and Microsoft