Two Sectors, Two Speeds

Woah, spooky. Almost like the The Wall Street Journal read my post from last Friday ! If you haven’t seen it I highly recommend reading Katherine Blunt and Jennifer Hiller on "Big Tech's Latest Obsession Is Finding Enough Energy ".

The past few days have been filled with old datacenter and energy sector friends and colleagues, and new connections as well, reaching out to share their electricity availability worries for the next decade. Those have been super interesting discussions, and I thought I would share my thoughts more broadly on how I see this going for the data center industry.

First, I want to be clear - this is a problem that will ultimately be solved. While data centers may be one of the most technologically advanced technologies ever invented, they are also schematically simple. Electrons go in the front, bits flip in the middle, and photons go out the back. We know how to do all three of those things. We just must do them on a kind of terrifying scale.

And we will because the stakes are too high to fail. The digitization of our economy is well underway with massive tailwinds and huge room to run. The capital is there as the tech sector is one of the most profitable enterprises humanity has ever built. The potential supply is almost infinite with solar, wind, hydro technologies on a tear and geothermal, SMR based fission, fusion all starting to show real deployment promise. There is no theoretical physical limit to how much transmission we can build; plus we know we can and will do it more efficiently. And storage innovation is racing ahead with grid scale batteries, inertia and stored heat systems, as well as green hydrogen all in the mix for making significant contributions.

But… it is going to be a brutal decade or two. Because you have two sectors operating at two fundamentally different speeds. The utility sector is not known for its rapid pace, and the tech sector is not known for its patience. And the tech sector’s data center expansion plans for the next decade far outstrip the utility sector’s ability to keep up.

How did we get here? I think it is because the success and scale of the cloud snuck up on even the cloud service providers. When I started at Microsoft in 2009 Azure was still called a ‘project’ (Project Red Dog for those of you who remember)! Who would have thought that in such a short time so much would have happened?

So what is going to happen? Well first, the tech sector has a little wiggle room by shifting to locations with abundant zero carbon energy for large scale centralized compute workloads (think training large language foundation models like the one used to create the banner image for this article). But that will only take them so far. From there they will need to either engage in the incredibly complex societal debate (policies, permitting, pricing, etc.) on public infrastructure development so that they can connect to a grid that is stable enough and large enough to serve their needs, or they will need to generate their own electricity. The problem with both of those options is they are slow, and they are expensive. Building national scale infrastructure in every country around the world doesn’t happen overnight and it sure ain't cheap. And it isn't clear that solving the science and engineering challenges required to realize the tech sector’s dreams of building data centers on top of nuclear fusion reactors will be any faster or cheaper.

And that is why I say access to electricity will be a rate limiting step to technology innovation over the next few decades.

The good news comes with a side of irony… which is that while the tech sector may be embarrassingly slow to understand its own basic dependence on the broader electricity sector, it is the most advanced in the world in understanding renewable energy technologies and markets. The reason it has achieved that level of expertise is because of its entirely voluntary sustainability and climate commitments - a often forgotten example of how being proactive on sustainability is about much much much more than simply 'doing the right thing'. And the way the sector will ultimately solve its electricity problems will ultimately reshape the world of electricity in the direction the world needs to travel to achieve a net zero carbon economy by 2050.

Although who knows… maybe by the time this all shakes out quantum computers will have fully arrived, nuclear fusion will be solved, and we will have access to infinite compute and infinite zero carbon energy and be back to wondering what we are going to do with all this abundance!