AI is taking off but it has a power problem. Tech companies are announcing massive new data centres but in many cases the electricity to run them is not there just yet. Joining us now is Andrejka Bernatova, CEO and Chairman of Dynamix Corporation III and Managing Partner at Dynamix Capital Partners. Andrejka, thank you so much for joining us today.
Thanks for having me Johnny.
AI companies have plenty of money. Why is electricity now the biggest thing holding them back and how is that changing where data centres get built?
We call it the California is coming closer to Texas problem — really electricity or power is the engine of the growth of data centres. It used to be GPUs. Right now it is really changing into the power problem, where you need to build the power infrastructure, which is extremely complex. It starts from permitting, it starts with local communities, then supply chain and building really large infrastructure where one piece is dependent on another. It is a multi-year problem and a multi-billion dollar problem.
You call some of these announced projects PowerPoint gigawatts. What do you mean by that?
It is one thing to announce a project and have aspiration to build gigawatts of power somewhere. The other piece is actually to execute on that. Permitting — we are seeing a lot of community pushback on some of those permits, including from states that are generally welcoming of power and energy infrastructure like Texas. It is one thing to have a project planned and another to execute. That is what we call PowerPoint gigawatts. What we have seen recently is developing from PowerPoint gigawatts not underpinned by contracts. Now at least we have moved into an environment where hyperscalers are willing to sign long-term contracts — which is the very first step to get the rest of the project off the ground.
As an investor, how do you tell a real project from one that only exists on paper?
Like any infrastructure — whether it was midstream, telecom, or highways — it needs to be underpinned by contracts or long-term cash flow viability. Step number one: is there a real contract in place? Step number two is execution — has the team building the power project built similar projects before? If you have a contract but it is not underpinned by execution capability, it is really not a contract of value. Shovel in the ground is a critical signal. I also think we are going to see not just big gigawatt announcements but a lot of decentralised activity — 100MW here, 200MW there. It is easier to permit, easier to execute, and easier to integrate into communities.
Where will all this electricity actually come from?
We strongly believe it is going to be a mix. Natural gas is the first step because it is the fastest and more reliable than renewables. But longer term, it will be mission critical for hyperscalers to have an energy mix — not just for clean energy purposes, which is expected, but from a resiliency perspective. When you look at history and democratic governments shutting down LNG pipelines and large projects, having an energy mix from fossil fuels to renewables will by definition ensure resiliency and protect against potential scrutiny down the line.
How do the US and Abu Dhabi approach this build-out differently?
It is fundamentally different. In America, the private sector drives a lot of the activity — it gets very fragmented and confusing, with a lot of hype and activity before the build-out actually starts. In places like Abu Dhabi, it is fascinating — it is a more controlled and systematic approach. You have the government involved, a lot of quasi-government money behind the build-outs. It is much more centralised and synchronised in the way the build-out happens.
Thank you so much for joining us today.
Thank you for having me.
