'When Opportunity Knocks' in AI. Nvidia, Meta & Instinct. ARD post 172

0:00 Welcome back AI curious folk to AI ramblings daily podcast with your host Michael Parrick your human AI guide. It's Monday. It's episode 172 again. Starting off with a bang. Today's theme when opportunity knocks, especially in all things AI, big companies, small companies. There are a lot of things going on and the companies involved that I want to talk about are grabbing for the brass ring. I'm going to go through each event as usual along with my take.

0:30 First up, of course, the kingmaker himself, Nvidia, Jensen Wong. They've announced just now $150 billion buyback, a record number for its stock. Remember, it's one of the few AI tech companies that's actually generating a lot of free cash as opposed to most of its peers and its customers who are spending not only the money they have in their pockets, but borrowing everything they can from debt or raising equity. Google raised over $85 billion to buy more Nvidia chips and also to develop its own TPU chips. And that's something all of its top customers, Nvidia's customers are doing. But Nvidia's valuation obviously has skyrocketed over the last 3 years over $5 trillion. Most valuable US company in the world. But on a relative valuation basis, it's in it's got a price to earnings ratio in the 20s. So they have a better use for the cash, which is buy back their stock, which the stock market likes. And that is a generally a good thing to do when those are the dynamics one finds itself.

1:36 So not not that much of a surprise. The surprise here is the big number and again a very market friendly action. While the company is very focused on the bottom up technical execution of it improving its platform every year Tik Tok to Black Rockck this year to Ver Rubin to Fineman you know on a one to twoyear cycle. So they're investing, they're acquiring, they're in they're buying companies, they're doing everything all of the above at a extraordinary execution pace. Absolutely focused on when opportunity knocks on all the doors in front of it. Let's move to the second item I want to talk about today, which is Meta reaching into the enterprise. I've talked a lot about Meta in the last few days about Muse, their AI agent product, which has some viral success across its three and a half billion user base. they've had millions of people download that and make made it the number one app on both the Apple store and increasingly on the Android Google stores etc. But there and I've written about this a lot but they are as I posited two or three years ago starting a new business line to sell their AI services to business customers and there's a big Wall Street Journal piece which I will have in the show notes. They hired CJ Desai the former CEO of DB to run it. He reports directly to the founder CEO Mark Zuckerberg. The new unit is going to bundle everything in Meta's AI toolbox, their viral Muse personal agent, their coding tools, their access to the top model, Muse Spark, which is at 1.3, their AI agents for businesses, which they're deploying across WhatsApp, which is their global successful service that is basically catering to a lot of small businesses on that side of the equation.

3:19 and they're spending more than a hundred billion actually over $140 billion on AI infrastructure this year alone. Investors want to see a return and Zuckerberg now having showed what he can do potentially with Muse AI agent on the consumer side is leaning in where the opportunity is knocking on the enterprise side obviously going up into a crowded space with Anthropics Claude Code with their upcoming trillion plus dollar IPO open Eyes Codeex with their trillion dollar plus IPO presumably next year SpaceX and its cursor acquisition for 60 billion and already an existing 1.5 5 trillion market cap company potentially to be merged at some point with Tesla by Elon which is the other $ 1.5 trillion company. So very crowded space on the enterprise side of the equation but Meta is doing what they can do given the capacity that they're building and that is actually the better thing for them to do if they can execute it but it's serious hard broad and deep execution to be done as they also obviously focus on the opportunity on the consumer side across WhatsApp across Instagram across Messenger across Facebook which are collectively are over three and a half billion people interacting with Meta's products daily and That's before we get to all the fancy things they're going to be able to do with over a 100 versions of these smart glasses coming this fall. So, a lot on Zuck's plate and they're moving fast to all of these opportunities. My overall take here, they've got a better than even shot at doing a lot of interesting things here, which is why the market is rewarding it. It's up over 25% since they announced the Muse AI agent success. And right now the tailwind is with them and opportunity is knocking and Meta is absolutely going after all of the knocks. So we will see how that transpires. Third thing I want to talk about is a small company year old company called Instinct AI. Most of you have not heard of it. They're a big viral success. They did a very smart strategy of having an invite only strategy to access its AI agent services available only to the who's who connected VCs and other muck mucks in Silicon Valley. It's a very positive strategy. They've but they have the product to boot. Their product actually is one that engenders in user trust early on. Over 40% of its users already give it a credit card within 3 weeks.

5:47 And that is a testimony to them being able to execute a complex service that works via messenger services like WhatsApp or Apple messages where you just ask things like chatbot but it does things for you but it earns your trust. It's a tough point to explain but it's a subtle one once you get access to the service. There was a useful podcast that I'll have in the show notes that the founder CEO Noah Shen did which is worth watching if you're curious about how they're doing this AI Asian thing. But they just raised over a billion dollars at a $10 billion valuation from the Huzu of VCOA benchmark and others. And this is a 4x kind of a growth from having raised $250 million at a $2.5 billion valuation just last month. So great execution on the business and then on the financing side of the equation. One of these things that you see in Silicon Valley once every few years and this young company with this young founder Noah Shin seems to have executed on all the opportunities knocking at their door and I wanted to highlight it because again AI agents are hot. We're talking about it in the context of meta huge company. This company's going to go up against them Google openi apple with Siri etc. that they're going to have to ramp up extraordinaryly fast in terms of their compute that they're going to need, which is why they've raised a billion dollars already when they have the opportunity. So definitely one to watch and if you're interested, just track them on the podcast etc. so you'll get a better sense of them and and try it when you get access to the service.

7:22 My overall take on all of this today in terms of this opportunity knock themes is the the core thing that we're seeing whether the companies are the largest ones out there in the AI land to the smallest ones the execution pace is something I have not seen continue at this pace year four if you count from the chat GPT moment in November 30th of 2022 or if you count it from the attention of all you need Google AI paper when transformers I invented back in 2017. So 9 years or four years, the relentless innovation that's going on at every level in terms of what you can do with these things is just changing dramatically. Even in this young company's instincts case that I just mentioned, they're growing 10% a day.

8:10 And that kind of gives you a sense that when things hit and you capture the lightning in a bottle, it's a lot of lightning to catch. So that's that's our theme on the main topic today. Let's move to the gadget AI section here. I want to talk about Tesla's Optimus robots. I've written a lot about this. Elon's been very focused on knocking on the opportunity door on robots. He actually did it even before he had a robot. His first Optimus robot was a guy in a suit on stage dancing and jumping a few years ago.

8:41 Since then, he's now ha he does have a production line that makes a few hundred robots a day, moving to thousands of robots a day. But the report that's out shows how that is even hitting speed bumps. There's supply chain issues in terms of getting all the components. There's complexity in terms of the robots hands. I've written a lot about this in prior posts. Again, all the the show notes will have the links for that if you want to take a look. They're running into those snags, but he's executing through them. But the key point is that the projections he has made, which is having millions of these things out in relatively short order, are going to take a little bit longer than anticipated. And that is something that Elon has never been shy on. And he's he's he's always the better early than never in terms of his projections on on things whether it's Optimus robots or self-driving cars. always sold potentially earlier, a few years earlier before their time. Same thing with data centers in space and all these other things. But that is Elon. He can execute, he can over time, he can, but more importantly, he can convince other people to support his attempts to answer the door when opportunity knocks earlier than than generally possible. But he does have the knack to raise the capital and the investor support and then build the execution capabilities to slowly but surely build these products. So that is the detail again there'll be a lot lot more in the show notes. Let's more move to the Q&A section of the today's program. Question one the instinct AI which I've which I'm very intrigued by. One question is what's my initial assessment? I've been using a lot of AI agents as you can imagine whether it's claude or open AI or meta etc. And instinct service I think is very deeply thought out and architected. They're hitting the issue of trust and privacy upfront. They're they're earning your trust slowly in terms of getting access to your services, your credit cards, etc. And then making sure that for the most part they deliver on those promises early. They're very focused on the delight aspect of the service. And their business model potentially is also very interesting and I'm intrigued by that.

10:55 The core thing is that they've already in this invite only member where they're barely have any members. They've already executed by their own account about a billion dollars in commercial activity. Half of it for travel related tasks, airline trips, hotels and the like transacted on its young platform. So they've had a lot of traction as I said before 10% a day very early. And that's it. It's it's been it's impressive to see not only how they've architected the thing, but how they built the their network to be able to support even this at this early stage this massively growing scale of platform demand.

11:31 Question two is what is the core challenge for them is the obvious one. They are competing with everybody and their uncle. Google, Meta, OpenAI, Apple, almost anyone you can think of wants to do AI agents at scale. Throw in Microsoft, throw in Amazon, everybody. And so they know this and they've got to run harder and they've got the same challenge that a young company got called Google that I remember meeting in ' 989 had. How do you go up against over a dozen providers of search when your search is better? You believe you believe it in your bones but you have to scale it much faster. Today that same challenge is there for this little company but the execution challenge is a thousand times more in terms of the capital needed in terms of the equipment needed in terms of the infrastructure needed etc. So it's going to be quite a task to watch this little company grow and execute to this in that in that wall of challenges but that's what they're doing when opportunity indeed is knocking on their door. So today's AI reset to zero piece is 1223. It's titled Frontier AI incidents are going from a trickle to a flood. It's of course of all this flood of issues that open an entropic and a whole host of other companies are finding with their models breaking out of their safeguards as they're being tested and what are some of the technical issues around that is what I'm discussing today in part one.

12:50 In the next two days there'll be a part two and three. Part two talks about the government regulatory impact of all of the these incidents and part three is talking about the hacking related things that are going on on a global state actors non-state actors etc. So it's a three-part piece a lot of detail and charts and graphs that I think you may find interesting if you have an interest in tracking this. So do watch out for today's post and the next two posts which I'll talk about tomorrow. Of course, tomorrow we'll be back, of course, with AI ramblings 173 and reset to 0ero 1224, part two of the series that I just talked about. I want to thank you again all for joining us, AI curious folk. Stay tuned.