Meta Platforms – LlamaCon I.

F8 is resurrected to shoot at OpenAI & Co.

• Meta Platform’s cancelled F8 developer conference has been resurrected as LlamaCon where Meta spent most of its time shooting at OpenAI, and setting out its case as to why it should be the platform for the AI era.
• Meta kicked off its event by disclosing that Llama has been downloaded 1.2bn times and then went on to launch two initiatives that aim to stop OpenAI dead in its tracks as well as ensure that Meta becomes the go to ecosystem of AI.
• These were:

• • First, Chatbot App: which is very similar to the ChatGPT app, Gemini app, DeepSeek app, Perplexity app and so on.
  • This is essentially nothing more than a web-based gateway to the cloud-based model which the user can use to ask questions and so on.
  • With foundation models rapidly commoditising, having a standalone app is not nearly enough because when it comes to this segment, OpenAI has all the users and everyone else has almost none.
  • Hence, one has to have something special or different to entice users to switch.
  • Meta aims to offer differentiation as its model will become personalised based on data that the user chooses to share with Llama (and Meta).
  • This is the kind of personalisation that Meta is extremely good at, having built the largest social media company in the world.
  • Assuming that users are willing to share data, this may offer a level of customisation that sets the app apart from the others, but that is a big if.
  • This is an attempt to encourage Facebook, Instagram and WhatsApp users to move from the ChatGPT app to Llama and steal some of OpenAI’s momentum.
  • Second, Developer API: which is by far the most important product and crucial to Meta’s strategy to conquer the emerging AI ecosystem.
  • This is a software development kit (SDK) that will allow developers to build, train and fine-tune their AI services on top of Llama as well as tools to evaluate the quality of the end product.
  • This is precisely what RFM research has been expecting (see here) and represents Meta’s attempt to become a control point in the emerging AI ecosystem.
  • The early release of Llama to open source has meant that it is already effectively the standard model that is used by developers outside of China, giving Meta a good start in the fight against OpenAI and the other closed-source offerings.
  • If developers now flock to Llama, then all of the cool new services will be developed on Llama first, meaning that ChatGPT will lose a significant amount of its appeal.
  • It will also help Llama to compete against Gemini where Google is keen to ensure that Android users predominantly use Gemini on Android devices.
  • The API is currently available as a limited preview and there is no word on pricing, but if Meta is smart, it will have priced it at 0% margin or lower to entice developers to give it a try.
• LlamaCon demonstrates the crucial juncture at which the fledgling AI ecosystem currently finds itself.
• There are large numbers of foundation models which cost large amounts of money to develop, and given that they are commoditising fast, only a few of them will ever earn a return on the investment that has been made.
• If developers are using Llama for developing their services, then Meta will ensure that its platform will be one of the winners.
• History has shown that owning the control point in any system is a license to print money, meaning that if successful, Meta will make a handsome return on the investments it is making, even if the method of monetisation is not clear at this stage.
• I expect that we will see Google make similar moves at Google I/O in May and OpenAI is also likely to redouble its efforts to win over developers and preserve the early lead that it has.
• This once again throws the spotlight on Apple, where Apple Intelligence has proved to be anything but intelligent leading to a lot of concern and substantial ridicule.
• Fortunately, this is not yet a problem as AI has yet to impact the choice of which device to purchase, but this may soon change.
• Apple has some time to fix this, but not an enormous amount, and the price of failure will be spectacular to behold.