Meta Platforms (META 2.34%) wants to give developers a new way to build generative artificial intelligence (AI) applications.
The social media company announced LLaMA 2 at Microsoft's (MSFT 1.97%) recent Inspire event. It's partnering with the IT giant's cloud computing service Azure to offer developers access to the open-source large language model, or LLM, which can serve as the backbone for new generative AI applications. Amazon's (AMZN +0.71%) users of Amazon Web Services can also use LLaMA 2 on Windows servers.
Meta isn't charging a licensing fee for its LLM. Instead, it expects to use the innovations developed by businesses using its technology to improve its own generative AI services and catch up with OpenAI and Alphabet's (GOOG +0.15%) (GOOGL +0.22%) Google.
Catching the competition
There's no doubt the two giants in generative AI are OpenAI (in which Microsoft has a significant investment) and Google. But smaller competitors are catching up quickly.
That's in large part thanks to Meta.
In February, Meta made Version 1 of LLaMA available for researchers. But the code leaked and developers took the foundational model and ran with it. In the span of about a month, new models were approaching levels that could compete with Google's Bard, despite the limited resources of independent investors.
The progress of open-source LLM development reportedly led one Google researcher to say, in a leaked internal memo, "We have no moat." With the release of LLaMA 2, Meta is throwing fuel on the fire of open-source LLM development and helping the open-source community catch up to the closed resources Google and OpenAI hold.
Indeed, LLaMA 2 is a meaningful threat to the profit potential for OpenAI or Google's Bard. Why would people pay to license their LLMs when they can get one that's just as good for free?
NASDAQ: META
Key Data Points
What does Meta get?
Meta has invested billions in developing and training LLaMA 2. CEO Mark Zuckerberg recognized the fact that simply using its LLM is costly. "A lot of the stuff is expensive," he said on Meta's fourth-quarter 2022 earnings call. That's not to mention the tens of billions his company spent on data centers focused on AI development.
So what does Meta gain from giving all of its hard work to developers to do as they wish?
Meta benefits from the improvements and innovations the open-source community can make. The company can take those innovations and feed them back into its own products. And AI is huge for Meta.
"We're exploring chat experiences in WhatsApp and Messenger, visual creation tools for posts in Facebook and Instagram and ads, and over time video and multi-modal experiences as well," Zuckerberg said on Meta's first-quarter 2023 earnings call. "I think this is literally going to touch every single one of our products and services in multiple ways," he added later.
Meta is already using generative artificial intelligence to help businesses develop better ad campaigns. It's also previewed tools like chatbots in WhatsApp and Messenger, which can help businesses scale interactions with customers. It's also working on photo editing tools for Instagram Stories users. Internally, the company is using AI to improve productivity.
In short, Meta is focused on building a superior foundational model that it can apply to its own products. The company already sees multiple paths to improve engagement and monetize chatbots. It just needs to get its model up to speed with OpenAI and Google.
LLaMA 2 is far from a death sentence for Microsoft or Google. The two may never release an open-source version of their LLMs. But even if LLaMA surpasses the capabilities of GPT or Bard, most instances will run on one of the big public clouds -- Microsoft, Google, or Amazon.
But the benefits to Meta can't be discounted. And investors bullish on the potential of AI should see Meta as a serious competitor in the space. Even with the massive run in share price this year, Meta's AI investments could still unlock a lot more value over the coming years.
