@bayesian_lad: Solid-State Transformers Part ...

Solid-State Transformers Part 5: $ENPH vs $SEDG vs Others

Last week, I discussed what SSTs are trying to solve for and how a few companies are looking to deliver a solution. Today, I want to compare them against each other (1/24)

View Tweet

Based on my research and some crowdsourcing from readers, I’ve identified 4 designs that we can assess:
1. Navitas $NVTS plug-and-play white-label tech
2. Eaton/Resilient Power
3. $SEDG SiC
4. $ENPH GaN Super Cluster (2/24)

It’s way too early to declare who the winner will be when it comes to providing the next generation of transformers, but there are already substantial differences in the designs such that different consumers may prefer different designs (3/24)

I’ll start with $NVTS as their role is somewhat different. Per my understanding, $NVTS is not building the SST. While they’re providing the chips to make SSTs work, they want to be providers to companies like Vertiv $VRT and Schneider (4/24)

In the ideal scenario for $NVTS, these companies would build their own SSTs using $NVTIS SiC chips and potentially GaN chips closer to the racks. $NVTS has mentioned Eaton $ETN as a potential customer, which makes the Eaton MVSST somewhat confusing (more soon) (5/24)

In this regard, $NVTS is competing more with companies like Infineon $IFNNY, the partner working with $SEDG (possibly $ENPH as well) .
An advantage of working with $NVTS is that these electrical equipment makers won’t need to build chip-making muscle in-house (6/24)

They can focus more on the switchgear, cooling, and other parts of the transformer. The downside is that until other power semi companies are in a similar position to $NVTS, $NVTS will have strong pricing power over their potential clients (7/24)

That being said, using $NVTS allows these blue chip companies to meet the new demands of next gen data centers, potentially allowing them to maintain their stronghold over the market. They’ve already built the relationships with customers (8/24)

Next, I’ll look at the $ETN solution. As far as I can tell, $ETN is the farthest along in its development. They’re far enough that they’re testing their MVSST with clients and even starting to quote prices (9/24)

Unfortunately, I have no idea what’s going on inside the SST. Gemini speculates that SiC is used. However, $ETN has the brand recognition to win these pilot projects. They’re only targeting late 2027 for orders, however, so it’s hard to say how close to ready they are (10/24)

Sticking with SiC, let’s look at $SEDG next. $SEDG is co-developing chips with Infineon. Infineon is doing most of the building, but $SEDG is actively leading the design and testing. The vision is impressive (11/24)

Unfortunately (and I’ll discuss this more in my $SEDG earnings recap), the latest earnings was light on concrete updates regarding the system. It could be that $SEDG is simply playing things close to the vest, but there isn’t much to dissect in their design yet (12/24)

That being said, if they can hit their target SST efficiency of 99% and overall system efficiency of 96 to 98%, I believe this would be superior to what $ETN is planning to offer. The solution will also be modular, with blocks available from 2 to 5 MW (13/24)

If $SEDG follows the same pattern as they have with their string inverters, these will likely be 1 SKU with software determining the capacity. If this is what ends up happening, buyers have a bit more flexibility to scale configurations (14/24)

Finally, let’s look at the $ENPH design before comparing what type of clients may prefer each. $ENPH is taking arguably the most novel approach. Each of their 1.25 MW SSTs are actually consists of 342 modules (15/24)

Skids can hold up to 4 transformers, scaling to 5 MW. Without going too deep into the math, $ENPH has redundancy built in; even if close to 10% of modules are down, workloads will still be supported (16/24)

Most important to the $ENPH pitch, however, is the efficiency gains through programmability. Enphase’s pitch is that with its product’s modularity, clients can match modules in-use to actual power demand (17/24)

This could raise overall system efficiency by reducing operating costs.
Alright, let’s move to which clients may prefer which solutions. IMO hyperscalers for now may favor an $ETN or $NVTS based solution (18/24)

The reputation $ETN has built is big asset. However, if $SEDG can demonstrate SST efficiency >99%, hyperscalers may be willing to take the risk and go with the newcomer. Higher efficiency translates to lower total cost, and that matters (19/24)

IMO neoclouds and co-los should favor the $ENPH solution. Any company that effectively rents resources out to others should want to minimize overhead whenever they can, and I believe the $ENPH solution will provide the most granular control (20/24)

Additionally, while most solutions will be modular to a degree, IMO $ENPH offers the most granular modularity, which makes upsizing, downsizing, and transferring easiest.
The biggest wildcard is total system efficiency for $ENPH (21/24)

If total system efficiency of the $ENPH solution, with savings coming from only needing ambient cooling and leveraging granular control to most closely match demand, can match or exceed the $SEDG and $ETN solutions, this could tilt the market in favor of $ENPH (22/24)

A double-edged sword for $ENPH is that even though its arguably the simplest system to maintain (easy to work with, readily available parts, etc.), the fact that so many parts are involved could make it harder to maintain (23/24)

This closes out my series on SSTs for the time being. I’ll provide updates as they come out. If you found this series useful, I’d appreciate you sharing this thread and the others. This is going to be a very exciting stroy. GLTA, and DYOR (24/24)