Nicholas Thompson’s Post

Here's the post: https://www.sequoiacap.com/article/ais-600b-question/

Nicholas Thompson always valuable insights, thank you – and AI and I approve of the hirsute aesthetic.

AI is definitely making waves, and the $600 billion investment in AI chips and data centers is a huge deal. To put it in perspective, it’s like the massive investments we saw in the telecom industry during the late 90s and early 2000s. Back then, billions were poured into laying fiber optic cables. Initially, there was a lot of skepticism because the returns were slow, and it seemed like an overinvestment. But over time, this infrastructure became the backbone of the internet boom, enabling the digital age we live in today. To contextualize $600 billion, think about it this way: it’s roughly equivalent to the GDP of countries like Switzerland. It’s also more than what massive companies like Apple or Amazon make in a year. This shows just how serious the commitment is towards AI. The key takeaway? While the AI sector is still in its early days with a big gap between spending and earnings, these investments could pave the way for groundbreaking innovations. So, while it seems like a massive spend now, it might just be setting the stage for the next big thing in tech. The only way to go from here is up! Let’s not forget how deeply integrated ai already is in our lives. There’s simply no turning back.

The amount of value created outside of the big tech companies is irrelevant to the big tech companies. You think they are going to lose money as an altruistic gesture to improve the world? 🤣 That is not why they are in this at all. Why would someone of your experience even think that??? Same thing with AGI - if the company that creates it can't control it and profit from it you think they are just going to gift it to the world? WTH. They'll terminate it just like any other non-profitable project once they realize they're never going to make ROI. Get real.

I am not nearly so smart, well connected and informed as you, but is AI investment really so different from standard capex investments that companies do constantly? Who would have said in 2001 'Intel is spending $100B on a new fab complex in Arizona that hasn't made any money yet' so they're a bad stock investment (for that reason)? No, ALL investments take time to spin up and recoup the money. Most of these returns are speculative - cannot be accurately calculated in advance. Why should a company lose value when they are actually gaining future increased earnings? That's shortsighted to me. I don't even see this is primarily as a race for first place in AGI. (Certainly that race does exist, but) companies are training their virtual assistants on different data with different rules and (I believe that) the resulting products will have different strengths and weaknesses. This variety of offerings should also undercut the theory of 'close to infinite' profit. Competition will undercut the market leader's profit margins, especially with a saturated market.

Not so fast #HealthTech #AI #Companies! You have to let #Nurses, #Physicians, #Pharmacists, & other #Licensed #Clinicians who live by an #OathofEthics lead every phase of the design, implementation, policies/procedures, ethics, & legal regulations of ANY #AI models that touch #PatientCare & #PatientData. #Licensed #Clinicians of all disciplines need to organize a group that demands to have involvement & approval of #HealthTech #AI products, services, policies/procedures, & regulations at facility level, state level, & federal level before they are implemented on #PatientCare. The business of #Healthcare & #HealthTech is already designed to maximize profit while being opaque, exploitative, inefficient, & dangerous for patients & frontline workers. National Nurses United American Nurses Association SEIU Healthcare SEIU UHW SEIU Doctors Council SEIU American Medical Association American Federation of State, County, and Municipal Employees Nicholas Thompson Ian Bremmer / GZERO Media Nita Farahany Dr. Wendy H. Wong Gabriela Ramos Eileen Donahoe JD, Ph.D Megan Metzger Karen Rommelfanger PhD Lucy Tournas, J.D. Amanda Zwarenstein Agnes McMahon Stephens

The market size was valued at around $62 billion in 2020 and is expected to grow at a compound annual growth rate (CAGR) of over 40%, potentially reaching $600 billion by the mid-2030s. According to a 2020 report by Flexera, 94% of enterprises use cloud services, with 87% taking a hybrid approach, using both public and private clouds. Some small businesses, especially those with specific regulatory, security, or performance requirements, may opt for on-premises data centers. These companies often have sensitive data or legacy systems that make cloud adoption less attractive. About 50% of small businesses (with less than 100 employees) use on-premises servers, though this number has been declining in favor of cloud services. By 2025, Gartner predicts that 80% of enterprises will have shut down their traditional data centers, but this largely applies to larger enterprises rather than small businesses. According to Statista, the edge computing market, which includes small-scale data centers, is projected to reach USD 15.7 billion by 2025. Numerous AI startups are emerging, focusing on niche areas and innovative applications of AI, contributing to the overall market growth.  

At the end of the day, he tells us essentially nothing. He says it could be good it could be bad some things will succeed some things will fail it's not a big problem but it could be a big problem but it will be okay in the end and so forth. His non-calculations based on somebody else's non-calculations ignore lots of things associated with costs, time frames, and so forth. It's just the sort of lightweight analysis and I see all the time from people who make lots of bad bets but have so much money that in the end they get the average which is pretty good.

This reminds me of the questions about broadband investment ROI at the Internet's outse. U.S. telecom carriers invested about $60 billion annually on average from the late 90s through the early 2010s. Did they fully capture the value created? No, of course not. The resulting businesses have market valuations in the trillions combined with many times the earnings of carriers. Nevertheless, carriers and data centers have done okay by differentiating their services and offering higher reliability or additional services for a premium. I'm sure that during the early industrial revolution, the math was similar (passenger vs. freight rates, but overall a small slice of the total value of goods and services that came to depend on non-horse based transportation). I think that infrastructure, however essential, ends up being treated like a commodity through overcapacity, competion or regulation (anti-trust). This despite the historical experience, which is that value generated through creative use of infrastructure ought to surpass the infrastructure investment by many multiples. I think its safe to say that AI firms have the data and analytics chops to see what is working early and find ways to move up the value chain quickly.

"The companies spending the money have loads of cash on hand, and they have ways to hedge their bets too." Irrelevant. AI revenues have to emerge or the companies will stop paying for Nvidia chips (and cloud services). I summarize the data in much more detail than this post does. https://www.linkedin.com/posts/dr-jeffrey-funk-a979435_datacenter-ai-technology-activity-7216032316601585664-RFXf?utm