Every quarter, I like to do a small update to give our readers a behind-the-scenes look at what is happening. Often, there is a big difference between what folks see publicly and the inner workings of STH, so I like to peel that back. This quarter has been all about a quick re-calibration and expanding as fast as we can.
Previous Updates
If you want to check out how this series has evolved, here are the links to the previous ones:
- STH 2019: Q1 – Q2 – Q3 – Q4
- STH 2020: Q1 – Q2 – Q3 – Q4
- STH 2021: Q1 – Q2 – Q3 – Q4
- STH 2022: Q1 – Q2 – Q3 – Q4
- STH 2023: Q1 – Q2 – Q3 – Q4
- STH 2024: Q1 – Q2 – Q3 – Q4
It is amazing how time flies.
STH Q1 2025 Letter from the Editor Hustle
The server industry has really changed in the last two quarters or so. Entering Q4 2024, we had two major headwinds. First, the AI server boom is making servers bigger. Second, Intel has largely pulled back its efforts in the industry. Those two points have had a dramatic effect on what we do on STH.
First, the AI server boom has meant that organizations have been spending more in the data center, but not in traditional compute. Servers that folks wanted us to test a decade ago were often $2500-$7500 and used at most 500-600W. Now, folks are asking that AI servers that are $100K+ get reviewed where each one can use upwards of 10kW.
That presents a logistical challenge on several fronts aside from the cost of running a server that uses that much power in one of our colocation labs. An example is we have seen systems get held up in customs at the end of 2024 with large customs fees and durations spanning two weeks in some cases. If you think about how long one of these servers stays “state-of-the-art” in an era of one year product cycles, two weeks spent doing nothing while sitting in customs is an eternity.
An alternative is to ship me to the servers as that has often become cheaper than shipping the system to STH. For systems in California, that is an easy day trip. If a system elsewhere, it is often a day or more eaten up in travel. For example, when we did the Quanta QCT QoolRack Liquid Cooling Intel Xeon Max Overview piece in Taiwan in 2023, that was a day to travel out, a day to travel back, and a day there. Of course, for a liquid-cooled rack, it is still much less expensive to ship me than the rack. I did a somewhat analogous piece last week on the east coast that took a flight out at 5:15AM on Monday morning and Sam and I did not arrive back until around 8PM on Wednesday. We were able to film somewhere cool, and it was a great trip, but that means it takes three days to film a piece instead of one if we do it in the studio.

The next challenge, related to that is even just liquid cooling. Everyone has different fittings and connectors. Even the fluid being used needs to be correct for the CDU, manifolds, hoses, connectors, coldplates and etc. in a system as the components can influence corrosion when used with different fluids. There is even more than that because if you ship a liquid cooled system via air to save time, then you have to deal with expansion of the air or liquid in the process plus whether to ship with or without fluid.
It has taken a lot of work to figure out how we will continue bringing higher-end systems content, but I think we have a good idea. The challenge is that every time we do an AI server system review, it is taking the time of doing multiple traditional server reviews. If we posted infrequently, then perhaps it would not matter. Posting every day, it has an impact.
On the Intel side, the pullback has been massive. A great example of this was the Q1 2025 Xeon 6 launch. We were the first to report in the Substack that the Xeon 6900E series would not be GA to the point you could select and order it in the average Dell PowerEdge, HPE ProLiant, and so forth. The Xeon 6700P series, Intel did not sample. Usually it is Michael Larabel of Phoronix and I that get the first calls in the industry since we drive the most traffic on websites to server CPU content by a longshot. Whereas normally we would have systems three weeks or more before launch, on this one, we did not. In fact, we still do not have chips from Intel for the Xeon 6700P series which is expected to be their high-volume part of this generation. The cuts have been deep enough that probably the highest ROI marketing in the industry, just getting chips out (usually they are QS when we get them), has not happened.
From a corporate perspective, some of this makes sense as Intel needs to conserve cash. Earlier this year I put AMD’s server market share in 2025 hitting 40% in our Substack. Just looking at the STH page view figures for AMD and Intel, there is an increasing possibility we will exit 2025 seeing higher numbers. If the trend continues, and Intel does not execute on the Xeon roadmap on time, then there is a chance that by the end of 2026 AMD goes over 50%. That feels more realistic in 2027, but in either case, that is going to be a big change in the market.
OEMs have figured this out and Intel has lost the position as the 800lb Gorilla in the server market. Five years ago, folks were nervous to show AMD models. Now, those same OEMs know that in the shift to more AMD EPYC market share, there is server market share that will shift.
Still, I thought that 2025 would be a down year for STH. That was totally wrong. We have been in a scramble to scale the team while at the same time dealing with the time per review going up.
Hiring in Scottsdale
It is time to scale the team a bit. This quarter Sam joined us as the primary camera person in studio. Some folks may have seen him show up in some B-roll like the recent Acemagic F3A video.
Some may have noticed that I have been chatting with Ryan Smith who did our NVIDIA GTC 2025 Keynote Live Coverage and a few other projects for us this year. At some point I need to transition my role and so we are looking at bringing some well-known folks into the fold.
At the same time, if you are in the Phoenix, Scottsdale, Mesa area and want to get involved, feel free to reach out.
We just need to scale.
The STH YouTube Update
This quarter, the growth of the STH YouTube channel picked up a bit as we published more. We reached 900K subscribers. Instead of doing a piece for every 100,000 (e.g. the 500,000 one) We are just going to wait until 1 million.

My best guess is that we will crest 1,000,000 sometime in July 2025 if the growth rate does not slow too much. At the same time, that means we are thinking about evolving the channel for a 1M+ subscriber audience. I still have a vision for what I want to do, and we are slowly moving in that direction. You are going to start seeing a lot more of that soon.
The goal of getting the new set finished in Q1 2025 did not happen. Hopefully it will in April as we have had materials trickling in. Alongside that, we have also been doing more road projects.
The Axautik Group Update and Substack
In Q1, we rolled out our new analyst arm, the Axautik Group. This is the “Analyst” part of our Analyst-Influencer-Other model. Our first PCIe Gen6 and CXL retimer research short was more popular than I expected.
The Axautik Group will produce content designed more for the financial communities. Some folks will be upset that AG content will be priced more for those communities. This is not a replacement for STH in any way. Instead, it will be designed to take much of what we learn by doing STH and turn it into formats we can monetize.
There is a question of whether this will be more of a subscription model or selling one-off reports. We started doing a Substack in August. That Substack has been growing really well and will become an important revenue source in 2025. It looks like that will be doing over 100K views/ month this quarter which is not bad.
If you can subscribe at work, this will be a bigger focus in 2025.
STH Labs: The Shorts Channel Update
As a 2024 project, we have the new STH Labs shorts channel here. That has been in a slow growth mode for the last few quarters.
I think we had a decent idea for this one and we are going to start posting mroe content there as the team in Scottsdale grows.
Subscribe to the STH’s Newsletter and YouTube
Did you know STH has a free weekly newsletter that comes out on Saturday with curated “Top 5” pieces from the week? We know you cannot visit every day, so we can deliver our picks for weekend reading directly to your inbox. Subscribing to the newsletter is easy. Here is the form.
We are not selling your e-mail addresses and MailChimp is managing everything at this point so you can subscribe and unsubscribe from the list as you want.
We are not promoting the newsletter via overlays and pop-ups. Those are very effective, but they are bad for readers. I do not like them, so as long as I have a say, we are not going to have newsletter signup overlays. I run STH as something I would want to visit daily even if I did not work on it.
Finally, subscribe to our YouTube and check it out here. Since that is a big focus at this point.
Subscribing to STH’s newsletter helps you see my favorite pieces of each week and a preview of the next week. Subscribing to the STH YouTube channel also helps us demonstrate our reach beyond just the website. Sometimes, doing things like showing power consumption or fan noise is easier than writing about it. This has also been something important for STH over the past few months. We have to demonstrate reach, and simple things like subscriber counts help explain reach.
Final Words
Thank you to all of our readers, Substack subscribers, clients, the STH team, and our advertisers for making everything possible. Just know, we are working hard behind the scenes to bring you cool content. In 2025 the goal is not to just do the same that we always have. Instead, I have been pushing to bring you things that I think are cool but maybe we have never had access to. In April 2025, you should see at least two pieces from that effort.
Thank you for the update, Patrick, and thank you for continuing to provide great material for All of us :)
I know there a lot of Gorillas in the market to try and tame, but–as a Home Labber–I really appreciate the little things.
Here’s to more *clink*
Given that push to megabuck AI racks needing massive facility support, I’d be curious to see another “in-house, colo or hyperscaler” cost-benefit article. (Admittedly I’m biased.)
@Patrick
How about you do something with the javascript MESS that keeps consistently wiping user comments before they are posted by refreshing the page?
Every half year or so I make the mistake of wanting to comment on an article … only to be reminded what a fool I am .. by my half-written post being summarily wiped. Again.
/yes, was just about to comment on the Crucial 128G piece … finished comment wiped, so will not bother again for a year or so/
Well done STH. Although I enjoy reading about AI servers, it is not something that helps me in my purchasing or needs. I miss the days you guys used to do more of the ‘small guy’ reviews of hardware for small datacenters/businesses/homelabs.
Don’t go all mega-bucks AI world and forget the small operations! We were after-all your primary audience, and probably still are?!