Deepwater Asset Management

Goldman Sachs and Sequoia Capital are skeptical that AI can generate enough value to warrant the $100's of billions being invested in the space. Both penned recent posts that suggest the economics don't make sense, but I think they're missing the bigger picture: AI is a Pascalian wager for megacap tech companies. If a megacap tech company bowed out of the AI race and AI ended up being as profound as bulls expect, that megacap tech company would be in Pascalian hell. Its competitors would eat up the biggest opportunity since the Internet. That company may even lose status as a megacap tech company. If a megacap bowed out of AI and was right, best case it saves $200b vs a $2t+ market cap. Who cares. Investors will have short memories for megacaps that invest heavily in AI even if AI never pans out (see META/metaverse). Investors will have short memories for those that disbelieve AI and end up being right. Pat on the back, now what can you do for me next. Investors will have painfully long memories for those who miss it. The rational strategy for all megacaps is to keeping investing aggressively into AI, and that's what I believe they'll do.

Do you think we're already in an AI bubble?

We’re giving AI to 75% of our customers at no extra charge. You read that right. Why? For many forward thinking companies and their employees, remote work has been a life changer. Companies can hire the best talent, regardless of the state where they live. Employees can spend more time with their families instead of commuting. But as a result, the founders, HR leaders, and accounting professionals at these companies are faced with a no-win choice: ◼️ hours spent deciphering confusing state compliance requirements, or ◼️ or thousands of dollars in fines.  _ When we introduced AI as part of our most expensive plan, we immediately saw how much stress it removed. However, that impact was limited to only a small fraction of our customers. It bothered us. So today, we’re adding AbstractOps AI Compliance Copilot to our most popular plan (Starter)... ** at no additional cost.** We’re able to do so without raising costs, because we operate incredibly lean, passing savings on to our customers. These customers, in turn, have graciously told their friends, and we’re so incredibly grateful. Head to the comments for my full thoughts on why we made this decision, and to help us spread the word.

Want an Apple Watch for your brain? Neurable is launching it this Spring. The MW75 Neuro headphones will track brain activity to optimize performance and focus just as the Apple Watch tracks our physical fitness. Founder/CEO Dr. Ramses Alcaide joined me on the latest The Deload podcast to talk about the new product as well as how Neurable's non-invasive BCI approach can compete with invasive BCI's like Neuralink and even how Neurable can make digital assistants like Siri smarter. Listen on The Deload. https://lnkd.in/eCC4stiU

Congrats to Dr. Ramses Alcaide and the Neurable team on a huge with for Neurable AI! More exciting things coming to shape the interface layer of the future soon.

If we get AGI by 2025 per Jack Kendall at Rain AI, it's not superintelligence that follows. It's infinite intelligence. The ability to spin up millions, even billions of instances of university level digital intelligences is like a real life infinite monkey theorem. Except the monkeys are smart machines. Put a billion intelligent machines on a hard unsolved problems and you're bound to find some solutions, even if by accident. Solved is solved. Infinite intelligence seems more likely than superintelligence in any estimable time frame, and it will certainly take a lot of compute. Makes sense why AMD raised its estimate of the AI acceleration market from $150 billion in 2027 to $400. They see the same AGI to infinite intelligence opportunity.

If you want to know where the AI world is going, and why we might get AGI sooner than many think, listen to Doug Clinton's interview with Rain AI's Jack Kendall.

Midjourney does $200m in ARR. CB Insights says that would make it a $10b company at 50x revs (in the ballpark of rumored rev multiples for OpenAI). What would Midjourney have to generate as a business for that to be a fundamentally great investment? Napkin math: - Internal studies we've done at Deepwater Asset Management suggest great long-term growth stories return the value of the company in FCF over 10 years. Some of the best do it even faster. - Simple math I ran: Midjourney would need to generate some where around $10b in revenue in year 10 (2033) with 30% FCF margins (roughly in-line with software greats) to clear that hurdle. - Assume a blended customer ARPU of $150 currently (base plan is $98/yr), the company has around 1.3m subscribers now. ChatGPT+ probably has close to 3 million by now. - Assume ARPU goes to $200 over the next 10 years, that would be ~50 million paid subs to Midjourney. A high bar, but Adobe has 10s of millions. More likely, Midjourney does more to serve enterprise and blended ARPU is much higher. - Is a FCF return of $10b in 10 years possible for Midjourney? Certainly possible. Hard to argue probable, but that's always the case for great growth companies. The valuation will never feel comfortable. If it does, it's probably not a great growth company. (Disclosure: Subscriber and fan of Midjourney). https://lnkd.in/eHeAxUYT

In 2024, there is a projected shortage of millions of H100 cards, which is +$100B worth of demand looking for GPUs. This will continue to push up prices, and the cost to acquire H100 cards is increasing yearly. Today, H100s easily go for $3 to $4 per hour per GPU, while earlier this year you could get H100s for half that. Whether a company goes with one supplier or another tends to be financing terms, upfront or month-to-month. Affordability is a real problem, as this will be a supplier's market for years to come. At Hydra Host, we go direct and save you time and money on GPUs. We have a new supplier with hundreds of H100 80GB cards with IB available before the end of this year for essentially $2 per hour. Ping sales@hydrahost.com or sign up - https://lnkd.in/gj5QQA8K

Zuck called AI a new art form on the META earnings call last night. He’s right. AI is an art form. The combination of the human generated kernel of an idea and the machine generated deeper output yields art. The quality of the human artist's idea decides the quality of the machine's output. Learning to use AI well is like learning to paint. Anyone can pick up a paintbrush and put something on canvas. The artist knows how to make the work speak at a deeper level. Prompting with AI is the same. https://lnkd.in/exCcZZTG