Meta Joins the AI Arms Race, Targets OpenAI with Open-Source Offering

Meta Platforms, Inc., the corporate umbrella covering social media behemoths Facebook and Instagram, is entering the high-stakes battle of artificial intelligence (AI). An exclusive Wall Street Journal report suggests that the tech giant is in the advanced stages of developing an AI model designed to go toe-to-toe with OpenAI’s cutting-edge models.

According to insider sources, Meta’s new AI initiative aims to be “several times” more powerful than its earlier Llama 2 model, which was unveiled earlier this year. In terms of scale, Llama was built on 70 billion parameters, a far cry from OpenAI’s GPT-4, which is speculated to have around 1.5 trillion parameters.

But Meta isn’t stopping at mere parameter count. The company has further plans to make the model open-source, paving the way for third-party enterprises to craft high-level text generators, analytics engines, and more.

Silicon and Strategy

Powering AI of this magnitude isn’t merely a code-level challenge; it’s an infrastructure battle. To that end, Meta has ramped up its hardware capabilities, procuring Nvidia’s H100 semiconductor chips—currently among the most potent and sought-after in the market. These hardware acquisitions signal Meta’s preparation to begin training its large-scale language model (LLM) by early 2024, with an anticipated roll-out sometime next year. Notably, the release is expected to follow Google’s LLM Gemini.

Interestingly, Microsoft, which is a significant backer of OpenAI, had earlier partnered with Meta to make Llama 2 available on Azure, its cloud computing platform. However, for this next-gen AI model, Meta reportedly plans to leverage its in-house infrastructure, making it a self-reliant behemoth in the AI sector.

A Global Tug of War in AI Development

This announcement comes at a time of intensified competition among tech giants and national governments to dominate the future of high-level AI systems. The UK government, for instance, recently announced a $130 million investment in high-powered chips aimed at AI development. Meanwhile, in China, new AI legislation has been enacted, and more than 70 AI models have been launched, according to Robin Li, the CEO of Baidu.

In summary, Meta’s ambitious foray into the next frontier of AI development marks a new chapter in the ongoing saga of AI competition. With its intent to create an open-source system that outperforms its predecessors and rivals, Meta is positioning itself as a formidable player in a global landscape where the race for AI supremacy is heating up like never before.