AI: Meta keeping up with the 'AI Joneses'. RTZ #706

Meta accelerated its ‘pedal to the metal’ on open source Llama AI models in this AI Tech Wave, with its first developer oriented conference. It’s notable given the DeepSeek from China competition on open source AI chatbot and reasoning/agentic models. Again highlighting the fast moving and accelerating nature of this LLM AI scaling race, especially this year.

The Information summarizes it all well in “Meta Plays Catch-Up, But Is It Working?”

“Meta Platforms kicked off its first-ever LlamaCon, a conference for AI developers, on Tuesday, unveiling a fleet of new products. They included a standalone smartphone app for its Meta AI assistant and a preview of an application programming interface for developers to access its large language model Llama. The releases put Meta on par with its competitors, but the question remains: What took so long?”

“After all, Meta had been working on its standalone app for Meta AI for at least eight months—two months longer than it took to build Twitter rival Threads. And to get more businesses using Llama, creating a developer API is pretty basic. (We scooped the company’s plans for the app and its API last summer and earlier this month.)”

“Rivals like OpenAI, Anthropic, Google, xAI and Mistral have long had both chatbot apps for consumers and APIs for businesses.”

Meta is addressing the API access piece in particular:

“With these releases, Meta is trying to catch up with its competitors by playing both the consumer and developer game, while not breaking particularly new ground in either arena.”

There are still some missing key AI pieces for developers.

“It’s noteworthy, also, that Meta didn’t release a reasoning version of Llama on Tuesday, as some developers expected it to. The company had planned to release that version several weeks ago, and then pushed back its release date, according to two people familiar with the matter. The reason for the delay isn’t known but Meta previously delayed other versions of Llama 4 over performance problems during its development.”

“Of course, many of Meta’s rivals have also already released reasoning models, which take more time to “think” before responding to a query.”

“We have followed the same [product development] process over the years,” a spokesperson for Meta said in a statement. “First we build a great consumer experience, scale it to be large—usually a billion people or more, and then once it’s at scale we start focusing on monetization and building out the business around it.”

“The most innovative feature from Tuesday’s releases is in a sector that Meta knows best: social. The standalone app for Meta AI includes a social feed where people can share how they are using artificial intelligence. (Be careful not to share whatever deep, dark secrets you’re telling Meta AI to your feed!)”

And Meta will have to balance user features, ad monetization, and of course privacy concerns. And keep up the AI capex investments with a little help from Nvidia and others.

“But to hear Meta CEO Mark Zuckerberg tell it, all of this competition vindicates the company and its open-source approach to AI. “The reality—and part of the value around open-source—is that you can mix and match,” he said on stage Tuesday. “If another model like DeepSeek or [Alibaba’s] Qwen is better now, as a developer … you just get this increasing mix, to pick the parts you want from an increasing number of open-source models.”

Investors are of course likely to be critical judges.

“Of course, for investors, that might not be confidence-inspiring, given the company has spent tens of billions of dollars on AI!”.

Microsoft CEO Satya Nadella stepped into sing Meta’s Llama model praises, despite the ongoing ‘frenemies’ competition amongst the key AI companies. Meta also reported strong results for the quarter.

All in all, Meta is working hard this AI Tech Wave to keep up with the AI Joneses, both in the US and China. In particular this conference emphasizes Meta’s new focus on the enterprise AI market I’ve discussed at length. Stay tuned.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)