3J·

Acute AI bottleneck #4: Energy/Grid ⚡

My last bottleneck posts were about HBM/Memory, Power & Cooling and Advanced Packaging. All three topics show that the critical points in the AI stack are shifting further and further into the physical infrastructure.

Today it's all about energy/grid.

Modern AI systems not only require computing power, but also enormous amounts of of stable energy. Several bottlenecks are now emerging in parallel. What I find particularly remarkable is that many investors think of energy almost exclusively in terms of electricity production. The actual bottleneck is spread across several levels of the infrastructure.

Level 1: Generation

This is where the energy itself is generated. For me $GEV (-0,11 %) (GE Vernova), $RR. (-1,11 %) (Rolls-Royce Holdings) and $BE (+0,34 %) (Bloom Energy) are exciting examples. The bottlenecks here are mainly in base load, flexible generation and local energy supply around new AI and data center load profiles. And Rolls-Royce could also become a player in Small Modular Reactors (SMR).

Level 2: Transmission

This is about being able to transport large amounts of energy at all. This sector is currently almost the biggest physical grid bottleneck. I find it particularly relevant here $PWR (+0,02 %) (Quanta Services), $PRY (+1,15 %) (Prysmian) and $HPS.A (+0 %) (Hammond Power Solutions). Typical bottlenecks are HVDC, high voltage, transformers and the massive expansion of the grid infrastructure.

Level 3: Distribution

Ultimately, the energy must be available locally, for industry, cities, data centers or critical infrastructure. There I look above all at $POWL (-1,03 %) (Powell Industries), $VICR (+6,18 %) (Vicor) and $MPWR (-0,28 %) (Monolithic Power Systems). The problems here are increasingly arising with power delivery, medium voltage, AI rack power and stable local grids.

Level 4: Implementation

This area is often underestimated. Because even if technology, capital and planning are available, projects still have to be physically implemented in the end.

That's why I think $FIX (+0,73 %) (Comfort Systems USA), $EME (-0,14 %) (EMCOR Group) and $STRL (+0,96 %) (Sterling Infrastructure) interesting. The bottlenecks here often lie in EPC capacities (plant construction), skilled workers, construction speed and the actual realization of large infrastructure projects.

Energy/Grid is currently so exciting for me because the bottleneck is not in a single place, but runs through the entire energy chain. but runs through the entire energy chain. That's why I consider this topic to be one of the most important bottlenecks around AI infrastructure at the moment.

#wikifolio
#scalelimits

7 Commentaires

Oliver Kepka@ScaleLimits

Outlook 🔭: Tomorrow I will write about what I see as the first FUTURE bottleneck that is looming large on the horizon: Test and Metrology. This is about zero tolerance, which concerns errors in the system and is due to the increasing complexity of components (e.g. chips).

•

Tenbagger2024@Tenbagger2024

Thank you dear, there are a few stocks in the portfolio that are already performing well. A few on the watch, of which there will still be company presentations. I already introduced Vicor a few weeks ago and also bought it

•

Oliver Kepka@ScaleLimits

@Tenbagger2024 Thank you for your feedback 🙏🏻 I think these are stocks that will also perform well in the future. I hold all of them in my Next Gen Technologies portfolio - most of them are also part of my three tiered bottleneck wikifolios.

••

Tenbagger2024@Tenbagger2024

@ScaleLimits $AEIS I also consider the sector to be very exciting, but also no longer so favorable

•

Oliver Kepka@ScaleLimits

@Tenbagger2024 Excellent choice. I hold the title myself, it fits perfectly into category 3 Distribution. $AEIS is not cheap. But possibly still early in the current infrastructure and AI power cycle and thus a medium to long term profiteer.

•

pivot_code@pivot_code

Thanks for the preparation!

Question: How do you get started here? :D

At the current valuations, I find it difficult to get into most stocks as long as there are no massive setbacks.

I actually see "only" $RR. $PRY and $EME as suitable, or am I missing something?

•

Oliver Kepka@ScaleLimits

@pivot_code Thank you 🙂 for your comment.

I can understand the content. At first glance, many titles in this area now seem expensive.

For me, however, it is often more important whether a company is sitting on a structural bottleneck or is benefiting greatly from it. This is precisely why I am currently looking more at themes such as energy/grid.

If you wait too long for the perfect setback, you are often just chasing rising prices. Comfort Systems $FIX is a good example of this for me: the share has repeatedly looked "too expensive" for a number of years and yet continues to rise structurally because the underlying bottleneck has become greater than the valuation concerns.

Of course, this does not mean that timing is irrelevant. If an RSI looks completely overheated, e.g. well above 80, I would also be cautious. But in the case of acute bottlenecks, it is often more important for me to be invested at all than to catch the perfect entry point.

And yes: I find RR, Prysmian or EME absolutely understandable in this context 🙂

•