Haven’t seen as much positive reviews regarding an AI product as I have around $META Muse. (Early ChatGPT vibes) Once $META start actively promoting Muse inside its networks I expect Muse to be the first AI assistant to reach 1B users.

The narrative on $META is about to change. 1. $META really surprised us with Spark 1.3, as it is a highly capable model on par with many even frontier models, while it is still not $META's big Watermelon model (coming soon according to Zuck and Alex's comments), which I expect to be significantly better than Spark 1.3. 2. The cadence of shipping models on which $META is working shows us that they set the foundation well and are now climbing on that ladder with compute. Muse Spark 1.2. was released just 1 month ago, and Spark 1.3 is much better than 1.2. More specifically, I think the ladder and the latest improvements of Spark 1.3 show that $META has figured out RL training well, which makes sense given $META's vast library of data, employees, and Wang, who knows the importance of data labeling from Scale AI. Given this cadence, I expect Watermelon to be shipped in a month or less. It could be that, in a few hours, as Astra might come out, Spark 1.3 won't look like a frontier model anymore. $META has no doubt leapfrogged a big gap it had and is coming closer and closer to the frontier. 3. $META has decided to go Scorched Earth Strategy with pricing, as they are one of the rare companies that can afford it given they have excess compute that doesn't (yet) have an outside buyer for it. If AI model labs get 80% gross margin, $META is fine with 50% or less, as this is not their core business and their main goal is to drive adoption. The scary part about $META for any AI lab is its AI data center footprint, which is huge (I wrote about it in a recent article on my newsletter). 4. $META can become a problem for frontier AI labs, not just because of competitive and pricing pressure on AI models, but as they might switch internal usage from Anthropic/OpenAI to their own models, and $META is one of the biggest token consumers with these labs. If $META has an equally capable model not paying 80% gross margin to AI labs because they use an internal model, it is huge, as $META for internal use-cases unlocks a lot more cheaper tokens. 5. At $META's valuation, 18x P/E, the market has priced in an almost zero percent chance that $META is one of the frontier AI labs. In fact, the market is even assigning negative value to $META's data center footprint. Given the recent trajectory of Spark 1.3. and given the amount of compute $META has the chance of them being a frontier AI lab is far higher than zero, and the market will start to price that. 6. Because $META itself internally develops frontier AI models that they can use for their core social media and messaging products and not get cut off in terms of access to an AI model, the competitive moat of $META's core social media and messaging assets increases drastically. If it was hard to compete with $META on social/messaging so far, it will only get harder. With owning the AI model, $META also controls its future margin on its core business.