543: The Nvidia Edition, Jensen's Peanut Butter GPUs, Anduril, Microsoft, Semi Export Controls, Protein, Fentanyl, Anthropic, Helsing, Tyler Cowen on AI Growth, and Moana 2

"a hidden signal waiting to be understood and harnessed"

Essentially, all models are wrong, but some are useful.

—George Box

There’s plenty of pie for everyone.

😒🏆😡👫 If you find yourself feeling jealous of someone, treat it like training a puppy: Stop, roll up a newspaper, and hit yourself on the nose.

But jealousy and envy aren't just unhelpful emotions.

There’s an opportunity there, a hidden signal waiting to be understood and harnessed by reframing/redirecting it into inspiration and motivation.

Jealously only drags you down, both in how you feel, and how you’re likely to act —because our actions flow from how we see the world. Think about it: Why aren't the pessimists the ones fixing the problems that they claim to see so clearly?

Being happy for others who are doing well and inspired by them is the useful side of that coin.

Turn envy into energy by celebrating others' successes and letting them inspire you to do better and reach higher.

There’s plenty of pie out there for everyone 🥧 — And we can create a lot more pie! The world is nowhere near its pie-creation capacity! 🥧🥧🥧🥧 — so why limit ourselves with zero-sum thinking? It’s self-defeating AND makes us miserable in the process.

Now don't get me wrong. I'm not saying that everything, everywhere is non-zero sum. But you have agency — you can choose to get out of lose-lose games and find win-win ones to play. The key is recognizing which is which and *actively* seeking out positive-sum opportunities instead.

It may not be easy, it may take time and planning and learning new skills, but it's worth every effort!

BTW, both photos above are fake pie. I made them with Midjourney. The model is getting really good at photorealism 📸

🏄‍♂️ To my delight, I’ve heard from several people that they bought balance boards after reading about my adventures in basement fitness.

I never thought I’d become a balance board influencer, but I’m glad — it’s such a fun thing to do, and the neurological benefits are a nice bonus.

A few supporters mentioned in the private Discord that the exact model I bought is available in the US too (when I first checked, I couldn’t find it). Here it is if you’re interested.

🛀💭 📚🎞️ 🎤 It’s a little strange to think that every Classic™️ — song, film, book, TV show, play, etc — once didn’t exist.

I mean, when you think about it, it’s obvious and logical.

Yet somehow it’s hard to imagine the world without some of these. They feel inevitable and timeless, as permanent as mountains 🏔️

And yet, someone sat down with a blank page and came up with them.

📺 🏯🎌 I started watching the TV series ‘Shōgun’ (2024, Hulu/FX).

Just three episodes in, and I'm already finding it so exceptional that I feel confident recommending it (be warned, the first episode has a really intense torture scene).

I’ll write more when I’ve seen the whole thing. I also added the James Clavell book to my ‘to read’ list (in Margins, of course).

---

🏦 💰 Business & Investing 💳 💴

🔌⚡️🇺🇸 U.S. Data Center Energy Usage Projections to 2028 📈🪫

This graph comes from a report by the US DOE and a few other US agencies.

It attempts to estimate datacenter electricity needs all the way to 2028 — no small challenge given the massive uncertainty created by the AI revolution, the accelerating electrification of transportation (EVs! Their growth may have temporarily slowed, but remains substantial and could surge again), the onshoring of some manufacturing, and the transition to more electric heating in buildings (Heat pumps! One of humanity’s greatest inventions — you wouldn’t have A/C without ‘em ).

Here’s what they have to say:

U.S. data center annual energy use remained stable between 2014–2016 at about 60 TWh, continuing a minimal growth trend observed since about 2010. In 2017, the overall server installed base started growing and Graphic Processing Unit (GPU)-accelerated servers for artificial intelligence (AI) became a significant enough portion of the data center server stock that total data center electricity use began to increase again, such that by 2018 data centers consumed about 76 TWh, representing 1.9% of total annual U.S. electricity consumption. U.S. data center energy use has continued to grow at an increasing rate, reaching 176 TWh by 2023, representing 4.4% of total U.S. electricity consumption.

With significant changes observed in the data center sector in recent years, owing to the rapid emergence of AI hardware, total data center energy use after 2023 is presented as a range to reflect various scenarios. These scenarios capture ranges of future equipment shipments and operational practices, as well as variations in cooling energy use. The equipment variations are based on the assumed number of GPUs shipped each year, which depends on the future GPU demand and the ability of manufacturers to meet those demands. Average operational practices for GPU-accelerated servers represent how much computational power, and how often AI hardware in the installed base is used, to meet AI workload demand. Cooling energy use variations are based on scenarios in cooling system selection type and efficiency of those cooling systems, such as shifting to liquid base cooling or moving away from evaporative cooling. 

Together, the scenario variations provide a range of total data center energy estimates, with the low and high end of roughly 325 and 580 TWh in 2028, as shown in Figure ES-1. Assuming an average capacity utilization rate of 50%, this annual energy use range would translate to a total power demand for data centers between 74 and 132 GW. This annual energy use also represents 6.7% to 12.0% of total U.S. electricity consumption forecasted for 2028.

The US electricity market is big! Going from 4.4% to a potential 12% — maybe even more if new developments occur that make demand accelerate faster than forecast — is A BIG DEAL.

“total power demand for data centers between 74 and 132 GW [in 2028]”

How many AP1000 is that? It’ll likely mean a lot of natural gas plants…

• 7% CAGR from 2014 to 2018 🐢

• 18% CAGR from 2018 to 2023 🐇

• And maybe 13%-27% CAGR from 2023-2028 🏎️

Clearly, they just put wide uncertainty bands around 18%… this is how sophisticated our forecasting is!

Time to build (a bunch of nuclear reactors)! 🏗️⚛️🏗️⚛️🏗️⚛️🏗️⚛️🏗️⚛️🏗️⚛️

🤖🤖🤖 Spreading the Silicon: Nvidia's GPU Allocation Strategy (And Why Microsoft Got More) 🤖🤖🤖

Guess who was at the top of Jensen’s Xmas card list in 2024? 🎄