Get live statistics and analysis of Hot Aisle's profile on X / Twitter

AMD already makes the top CPUs and FPGAs on the planet. The concept that they can’t make top GPUs, seems absurd.

In case it’s not obvious, I’m probably the most bullish person on $AMD AI outside of Dr. Su herself. I could’ve taken the (relatively) easy route and joined CoreWeave years before they went public. I’d probably be wealthy beyond anything I can imagine right now. Hell, I even pushed a close friend to go work there, and he’s loving it. But I didn’t. I chose the hardest possible path. I bet my entire business, years of my life, and an absurd amount of stress on this role. I went all-in on $AMD almost a year before $AMD was even talking about a real public AI story. I can’t count how many times people called me crazy, asking “why?” Why? Because this isn’t about money or fame. I don’t need either. I’ve been on the internet since 1991, this has been my entire life. I’ve seen every cycle, every boom, every collapse. And I’m genuinely worried about the long-term health of AI if the entire industry is chained to a single provider for both hardware and software. I believe strongly that this is a transformational technology for humanity, and monopolizing it is a globally dangerous proposition. The truth is simple: $AMD is our only real shot at preventing that. There is no “third place.” It’s you. You’re the hope. Take it seriously. So, if I’m ever openly critical of $AMD or anyone else in this space, it’s because I’m focused on where things are actually headed. There’s a reason $AMD is in second place (many reasons, honestly), and you don’t get better by pretending everything is perfect. If different choices had been made years ago, we likely wouldn’t be in the position we’re in now. Stop acting like everything is fine. Rip the bandaid off. Something has to change. Long live $AMD.

On the news today: AMD isn't dead, ARM is dead.

❤️‍🔥

AMD already has an APU design. Nvidia wants one, but not on ARM. Imagine a box of 8 x86/MI500 APUs. It is called a FLOPHOUSE!

What I learned in mining with 150,000 GPUs is that they are all snowflakes. No two are the same and there are wide variations between individual chips, even with the same exact settings. We called this the “silicon lottery.” I automated the tuning of these chips, which involved tuning the memory, clock, and power settings in precise ways in order to get best results. When they publish numbers, it is from the snowflakes that look like unicorns. We called this “Jensen math.”

"No VMs. No CPUs." Wait a second. Did you guys seriously buy 1,024 servers without decent CPUs? They can't run VMs because there aren't CPUs? What is this nonsense?

We offered free access to our mi300x boxes and he called us an AMD shill and blocked us. ¯\_(ツ)_/¯

AMD already makes the top CPUs and FPGAs on the planet. The concept that they can’t make top GPUs, seems absurd.

On the news today: AMD isn't dead, ARM is dead.

"No VMs. No CPUs." Wait a second. Did you guys seriously buy 1,024 servers without decent CPUs? They can't run VMs because there aren't CPUs? What is this nonsense?

We offered free access to our mi300x boxes and he called us an AMD shill and blocked us. ¯\_(ツ)_/¯

❤️‍🔥

AMD already has an APU design. Nvidia wants one, but not on ARM. Imagine a box of 8 x86/MI500 APUs. It is called a FLOPHOUSE!

In case it’s not obvious, I’m probably the most bullish person on $AMD AI outside of Dr. Su herself. I could’ve taken the (relatively) easy route and joined CoreWeave years before they went public. I’d probably be wealthy beyond anything I can imagine right now. Hell, I even pushed a close friend to go work there, and he’s loving it. But I didn’t. I chose the hardest possible path. I bet my entire business, years of my life, and an absurd amount of stress on this role. I went all-in on $AMD almost a year before $AMD was even talking about a real public AI story. I can’t count how many times people called me crazy, asking “why?” Why? Because this isn’t about money or fame. I don’t need either. I’ve been on the internet since 1991, this has been my entire life. I’ve seen every cycle, every boom, every collapse. And I’m genuinely worried about the long-term health of AI if the entire industry is chained to a single provider for both hardware and software. I believe strongly that this is a transformational technology for humanity, and monopolizing it is a globally dangerous proposition. The truth is simple: $AMD is our only real shot at preventing that. There is no “third place.” It’s you. You’re the hope. Take it seriously. So, if I’m ever openly critical of $AMD or anyone else in this space, it’s because I’m focused on where things are actually headed. There’s a reason $AMD is in second place (many reasons, honestly), and you don’t get better by pretending everything is perfect. If different choices had been made years ago, we likely wouldn’t be in the position we’re in now. Stop acting like everything is fine. Rip the bandaid off. Something has to change. Long live $AMD.

😭

Smoking Guns Now that Dylan, Semi, and Tensorwave have all blocked me here, I feel free to share what I see as the smoking gun. If you don't have anything to hide, you don't need to block people. This goes back to a conversation in January between Oren (who worked for Semi when the analysis was published) and Jordan Nanos (who didn’t work for Semi then, but does now, on v2.0). v1.0 came out in March, so there is no way Oren's views changed. SOC2 is a checkbox It makes one thing obvious: Semi never actually cared about SOC2. Yet in their report, we were the only other AMD specific NeoCloud flagged as “underperforming,” front and center. That wasn’t an accident, it was done to damage our reputation. The article went so far as to suggest that AMD only partner with companies that have SOC2 in an attempt to get us removed. TW already had SOC2, so that became their “advantage” over us. Despite Oren acknowledging that SOC2 doesn't prove anything. Surprise What Dylan (and probably TW) didn’t expect was for us to go out and get SOC2, and to do it faster than anyone thought possible. He refused to acknowledge when we got it and even moved the goal posts trying to suggest it was more than that (the screenshot below proves it was just SOC2). He also assumed I’d stay quiet like everyone else in this industry, where fear keeps most from speaking up. But I keep my chin high, I've got nothing to fear, and I won’t play along with their games. Grifters in this industry When you look at how they’ve treated us, and others in this space, it’s impossible to see Dylan as merely careless. He’s actively harmful. The final straw came when Dylan and Semi were hacked multiple times last week, potentially exposing all of their DMs and more. They push security standards on others while failing to meet them themselves, then try to sweep it under the rug. I’m not letting that slide. What people need to realize is that Dylan is a net negative for this industry. His whole playbook is engagement farming, throwing companies under the bus to juice clicks and in order to get paid for content. That’s not okay, and it’s not something we should be celebrating. There are plenty of other great analysist's with ethics. Working hard isn't an excuse for this behavior. ¯\_(ツ)_/¯ He’s scheduled to keynote at the upcoming PyTorch conference, and I’ve filed a formal complaint to have him replaced. In my view, there are far more deserving voices in this industry. He’s a false prophet and a grifter who preys on the community, and I know I’m not the only one who feels this way. I honestly don’t have much hope of success given his massive following compared to mine, but it’s important to me to at least stand up and try. TL;DR: We've got to stop idolizing jerks.

SemiAnalysis is acting like the cloud mafia. Everyone’s careful around them because they’re afraid of what will show up in the reports and backchannel conversations. I refused to play along, went on the offensive, and still landed Bronze. It just shows what a farce all of this is. At the end of the day, I’m here for my customers, not to impress analysts. Our customers are happy, that’s the only metric that counts. If that is a niche, so be it. Every company on the list below is filled with humans that are grinding to build something real. To state "Not recommended" -- they don’t deserve this treatment. It is beyond rude and unfortunate that this is happening. Dylan should be shunned from the industry, not applauded.

Just passed 1500 followers. So grateful! It has been all organic so far. Next goal is 2000. Any advice on how to get there? Fingers crossed that this coming week we should have a couple big announcements.