- Porter's Daily Journal
- Posts
- It’s Not A Bubble
Nvidia Is Cheaper Today Than It Was Before ChatGPT
This is Porter’s Daily Journal, a free e-letter from Porter & Co. that provides unfiltered insights on markets, the economy, and life to help readers become better investors. It includes weekday editions and two weekend editions… and is free to all subscribers.
Nearly half a trillion in capex… Nvidia’s growth is astronomical… Shares haven’t kept up with earnings… Huang’s Law outpaces Moore’s Law… It isn’t a bubble for Nvidia… Credit delinquencies soar… Bankruptcies rise… |
Table of Contents
It’s not a bubble.
Last year we published The Parallel Processing Revolution. This is the most important piece of research I’ve ever written. If you haven’t read it yet, I’d urge you to do so. And if you haven’t read it recently, I’d urge you to review it again now.
In the last three years, Microsoft (MSFT) has invested about $100 billion in 15 major new computer centers around the world and is rebuilding its existing Azure infrastructure, using parallel computing. Amazon (AMZN) has spent substantially more, about $200 billion, on 20 new computer centers and revamping the existing Amazon Web Services infrastructure. Google (GOOG), in part because it’s been using its own, proprietary chips, has “only” spent $75 billion on 12 new centers. And Meta (META) has spent roughly the same amount, while adding seven major new centers.
In total, just from these four companies, that’s almost half a trillion dollars! And believe it or not, investments in this new technology are accelerating. Microsoft plans to invest another $300 billion by 2030, to build more than 50 new centers each year. Amazon is being even more aggressive: it plans to double its compute capacity by 2027, while spending $400 billion by 2030. Google and Meta both say they will spend $300 billion through 2030. In total, this is an additional $1.5 trillion in compute capex through 2030.
The AI applications being developed (like ChatGPT, Grok, and Gemini) powerful data centers aren’t the only result of these huge investments. AI is merely one application of the ongoing parallel processing revolution. By 2030, there will be new applications that we can’t even imagine today – wholly autonomous transportation, human-like robots, unprecedented increases in productivity in virtually every industry, but most notably in education, finance, and medicine.