hist78
Well-known member
I like it too, after all it’s not my money. Those investors are essentially funding my wonderful learning experience.
I think Substrate is a $1 Billion Fraud: Part 1
- Thread starter Daniel Nenni
- Start date
I like it too, after all it’s not my money. Those investors are essentially funding my wonderful learning experience.
Xebec
Well-known member
Unfortunately, this type of situation is surprisingly common: https://www.techspot.com/news/107962-nvidia-rtx-5060-launch-erosion-independent-gpu-reviews.html
Reviewers are required to sign NDAs for NVidia GPUs (normal), but they're also given requirements on how the reviews should be conducted. For example, reviewers were told they were not allowed to compare the 50 series vs the previous gen 40 series... to avoid showing how small performance gains were this generation.
..
Nvidia also banned a few reviewers (later reversed after public outcry) for not posting positive reviews/strictly following this "guidance" about their products: www.reddit.com/r/hardware/comments/kbylt3/nvidia_apologizes_reverses_decision_to_ban
You're very right to point out the issues here; Substrate + Semianalysis feels a bit like this conflict.
I'd contend that Intel knows a thing or two about failure.
Kidding aside there is a massive target on the back of ASML right now, and if there was a quick path to 2nm without EUV China would be there already.
Last edited:
It beggers belief that people this naive are left in charge of millions of venture capital funds
EganVector
New member
Are Adam Neumann and Elizabeth Holmes on the board of directors for Substrate?
hist78
Well-known member
It’s common for companies or individuals to try to influence the tone and direction of reporting. But ultimately, the authors must put integrity above everything else.
Take the SemiAnalysis' Substrate report as an example. On one hand, they claim authority because they have access to Substrate’s internal information. On the other hand, they also can disclaim responsibility for any inaccuracies by pointing to NDAs signed by some members of the author team.
In the same report, they wrote:
Does this imply that Substrate has been paying SemiAnalysis for services since 2022? If so, that introduces a clear business relationship that readers deserve to understand up front.
I don’t expect the rigor of a PhD dissertation from this type of industry report, but at least we should know whether the analysis is an objective technical evaluation or something closer to marketing material.
MKWVentures
Moderator
Couple things on this:
As as consultant, I always had NDAs with Client. I had NDA clients who I was doing business development for. the writing in the NDA is that I cannot give out info the company does not want to give out. But I did sales pitches. Every employee of every company is in a similar situation. the key is to be honest and disclose what you can and cannot talk about. This is not uncommon for consultants or employees. Intel people say "our process is amazing and had leading edge leakage data. I cannot tell you the gate materials"
That said: Substrate is an idea that has not been fully checked out. Calling it a game changer is not correct. Calling it fraud is not correct (unless you have a lawyer and a proof).
If its real, lets discuss it at a lithography conference @Fred Chen can provide more technical analysis
hist78
Well-known member
I can't find any name for Substrate's board of directors so I can't make any comments.
Until recent years, I didn’t realize how out of touch a board of directors can sometimes be with the company they are supposed to know intimately.
The scandal of Elizabeth Holmes and Theranos really changed my view a lot. According to Google search, Theranos’ board of directors used to include:
Board of directors in October 2015. Before a corporate restructuring in late October 2015, the board included:
founder and CEO Elizabeth Holmes,
president and COO Sunny Balwani,
former CEO of Bechtel Group Riley P. Bechtel,
retired Marine Corps General James "Jim" Mattis,
former U.S. Secretaries of State George P. Shultz and Henry A. Kissinger,
former U.S. Secretary of Defense William J. Perry,
former U.S. Senator Sam Nunn,
retired U.S. Navy Admiral Gary Roughead.
former Wells Fargo CEO Richard Kovacevich,
former CDC director William H. Foege,
heart and lung transplant surgeon and former U.S. Senator Dr. William Frist.
Board of directors in May 2016:
Following the 2015 restructuring, the board was reduced to five members, and a new Board of Counselors was created, with Kissinger and Shultz moving to it. The resulting board consisted of
Elizabeth Holmes,
Riley P. Bechtel,
lawyer David Boies, William H. Foege, Richard Kovacevich,
General James "Jim" Mattis.
In May 2016, Fabrizio Bonanni joined the board, while Sunny Balwani resigned.
MKWVentures
Moderator
If every overhyped technology that was going to be a game changer was called fraud after failing to deliver, we would need about 1000 more prisons for fraud.
Boards have MUCH MUCH less control of a company and knowledge of operations than you think. Ask LBT
Any idea how much the Theranos board was paid? Money does sway opinions…
hist78
Well-known member
When I saw that Henry Kissinger was on Theranos’s board, I thought it must be worth a ton of cash. And he was one of the two former Secretary of State on the Theranos' board.
Xebec
Well-known member
I've been trying to get more useful information about Substrate.
There is an interview with their CEO dated 10/30/25 .. that is behind a paywall: https://stratechery.com/2025/an-int...out-building-a-disruptive-foundry-in-america/
The CEO tweets or retweets about once per month. The posts are pretty random and I did not see anything about Substrate, until very recently (4 days) where he reposted the Semianalysis article on Substrate.
Subtrate's X account has very little in the way of posts or replies.
It's not unusual for startups to be 'stealthy', but there just isn't much out there about Substrate. I can't find any information about any possible board of directors or find any interesting names attached to the company on the technical side.
I guess we'll have to wait and see if anything comes of this over the coming months/years..
There is an interview with their CEO dated 10/30/25 .. that is behind a paywall: https://stratechery.com/2025/an-int...out-building-a-disruptive-foundry-in-america/
The CEO tweets or retweets about once per month. The posts are pretty random and I did not see anything about Substrate, until very recently (4 days) where he reposted the Semianalysis article on Substrate.
Subtrate's X account has very little in the way of posts or replies.
It's not unusual for startups to be 'stealthy', but there just isn't much out there about Substrate. I can't find any information about any possible board of directors or find any interesting names attached to the company on the technical side.
I guess we'll have to wait and see if anything comes of this over the coming months/years..
EganVector
New member
I suggested those two as a joke, but clearly some people didn't get it and I'm really glad I didn't use Max Bialystock. MKWVentures wins the internet for today; truer words have not been said.If every overhyped technology that was going to be a game changer was called fraud after failing to deliver, we would need about 1000 more prisons for fraud.
Boards have MUCH MUCH less control of a company and knowledge of operations than you think. Ask LBT
This topic came up at an event I attended last night. It was deemed a fraud but we shall see. Dylan Patel's name came up as well. I hope he does not get tangled up in this.
One opinion of Dylan Patel rang true and this was from a person who has worked with him in the past. "Dylan is like AI, he googles around, attends conferences and gathers information. He does occasionally hallucinate but Dylan's LLM is growing."
I see this as a positive. I hope he does change the analyst game, it is long overdue. It seems like anybody can be an analyst these days. I guess they call them influencers now? When I register for conferences I just say media so I can get in for free.
One opinion of Dylan Patel rang true and this was from a person who has worked with him in the past. "Dylan is like AI, he googles around, attends conferences and gathers information. He does occasionally hallucinate but Dylan's LLM is growing."
I see this as a positive. I hope he does change the analyst game, it is long overdue. It seems like anybody can be an analyst these days. I guess they call them influencers now? When I register for conferences I just say media so I can get in for free.
Xebec
Well-known member
FWIW, In the financial industry, "anybody can be an analyst" has been the case for a very long time
If this new company aims to replace ASML and claims to work with TSMC or Intel, then TSMC or Intel could verify the legitimacy of its technology. However, since the company isn’t actually collaborating with anyone, there’s no way to verify whether its claims are true or not.
It has no US patent applications since 2017, based on assignees list here: https://patents.google.com/?q=(X-ra...+wafer+nm)+country:US+after:priority:20170101
IanCutress
Member
Just to fill in a bit here about Dylan's background, for those wondering.
He's a 29 year old whose business is ~$40m revenue, and growing. His take-home is at least 5m of that.
His academic background is finance. What specific area of finance I'm unsure, but in short he understands the markets and what the traders want to know.
He's been a moderator of the Intel subreddit since the age of 12, although hasn't done much with that in the last 5-7 years or so.
He spent a chunk of time before Semianalysis riding bikes with Charlie Demerjian (Semiaccurate) in the north. Charlie is a unique character, love him or hate him, but Charlie is just out for the truth. Charlie has a lot of friends and number of fingers in a number of pies.
Dylan's family also runs a handful of hotels, enabling him to stay at [well known name brand] for a 90% discount, enabling him to be a digital nomad for a decade.
Semianalysis was, for the first 2/3 years, a way for Dylan to post about the leaks he had access too. A good chunk were shitposts on purpose, but it got some notice, especially when he posted all about Blackwell the morning of the GTC announcement and they kicked him out of the event (or as-close-to-removal as possible). As far as I understand, he was asked/told to apologise to NVIDIA's CTO. Now he has regular 1-on-1s with Jensen a handful of times a year.
Around then, he did get his act together. I think his actual break-out moment was when he went to Hsinchu, and took a bike around the TSMC fab sites. I'm literally talking about the arterial roads around them, not actually inside TSMC owned property. He took a number of photos, posted them on social if I recall, and Taiwanese media went nuts. Some even said 'TSMC invites Top US Analyst Dylan Patel to latest Gigafab' or some other BS. But his name went far and wide, simply for a bike ride.
Nowadays, Semianalysis is all about creating AI accelerator and buildout models for wall street investors and companies in the industry looking to understand pricing and supply chain numbers. He built his team from 1 person to 4 to 12 to 25 and now it's around 40 people full time. How does he get the data? The same way most analysts do - tracking and conversations. He regularly advises investment funds and companies on what's going on in the industry, and they also share data with him and his team. I've seen some of the fees he's paid to present at investment fund retreats and such. I know for the AI datacenter build-out, they track ~1200 datacenters world wide, to the point where they hire drone photographers to get areal shots of facilities so they can literally count how many generators or batteries or anything that are being built. By building up from the base of these numbers, you can start to project the requirements, the hardware, the supply chain, and everything else needed around compute and AI. That's what's in the spreadsheets that Semianalysis sells. 'The most expensive spreadsheets in the world'.
Like I said, the company is about 40 people now. Dylan is the head and face, but the majority of the work is done by others. We are talking people with semiconductor backgrounds, with finance backgrounds. Imagine millennials and gen Z who have grown up around technology for the most part, and a number of seasoned veterans. A number of them (including Dylan) grew up reading my content and the like. They do have stupid internal contests, such as 'who can drive the most engagement on twitter with memes on the official account', but that's meant to be the 'casual' side.
Dylan and I are quite friendly. Happy to chat and have dinner and whatnot. He's not ever asked me to join his company, and knows I'll willingly call him out on his BS, and have done regularly.
But yeah, while Dylan is Semianalysis, he doesn't do all the work. He's one person of 40. Just like Jensen is one person in ~40k.
Attached is a slide from his recent CLUSTERMAX2.0 launch, listing all the employees. Feel free do your own research.
--
This substrate thing did twang the bullshit part of my brain. There's physics, previous research, and the economics of scale. The Semianalysis article did what every other article in the ecosystem seemed to do - get overly excited at the prospect of the potential of a future promise (and because 'MERICA!
'. It meant a lot of the physics and economics discussion seemed relegated to the wayside (or put behind the paywall). It blatantly missed obvious questions and didn't do any job of trying to answer them. I'm the perennial skeptic when people come with extraordinary claims, especially ones that I can run the mathematics on. Unfortunately I think the SA article was written by someone without that mindset, which is a shame because they have been pivoting to a more STEM attitude in a lot of the coverage.
I scripted and recorded a 12k word video about it, still editing. It'll be about an hour long - 30-35 minutes of it is simply going through chip design, where x-rays would work, the physics of what they do, before hitting the claims and economics. Those first 30-35 minutes will probably be super basic to anyone here, or 'no-duh' style of a high-level explainer, and the rest is basic math. Video should go up next week.
The CEO did reach out to me, about 10 minutes before my paid-for studio time was about to start. So there's a disclaimer in there that I'm due for a call. I'll likely go visit when I'm in town again for IEDM.
He's a 29 year old whose business is ~$40m revenue, and growing. His take-home is at least 5m of that.
His academic background is finance. What specific area of finance I'm unsure, but in short he understands the markets and what the traders want to know.
He's been a moderator of the Intel subreddit since the age of 12, although hasn't done much with that in the last 5-7 years or so.
He spent a chunk of time before Semianalysis riding bikes with Charlie Demerjian (Semiaccurate) in the north. Charlie is a unique character, love him or hate him, but Charlie is just out for the truth. Charlie has a lot of friends and number of fingers in a number of pies.
Dylan's family also runs a handful of hotels, enabling him to stay at [well known name brand] for a 90% discount, enabling him to be a digital nomad for a decade.
Semianalysis was, for the first 2/3 years, a way for Dylan to post about the leaks he had access too. A good chunk were shitposts on purpose, but it got some notice, especially when he posted all about Blackwell the morning of the GTC announcement and they kicked him out of the event (or as-close-to-removal as possible). As far as I understand, he was asked/told to apologise to NVIDIA's CTO. Now he has regular 1-on-1s with Jensen a handful of times a year.
Around then, he did get his act together. I think his actual break-out moment was when he went to Hsinchu, and took a bike around the TSMC fab sites. I'm literally talking about the arterial roads around them, not actually inside TSMC owned property. He took a number of photos, posted them on social if I recall, and Taiwanese media went nuts. Some even said 'TSMC invites Top US Analyst Dylan Patel to latest Gigafab' or some other BS. But his name went far and wide, simply for a bike ride.
Nowadays, Semianalysis is all about creating AI accelerator and buildout models for wall street investors and companies in the industry looking to understand pricing and supply chain numbers. He built his team from 1 person to 4 to 12 to 25 and now it's around 40 people full time. How does he get the data? The same way most analysts do - tracking and conversations. He regularly advises investment funds and companies on what's going on in the industry, and they also share data with him and his team. I've seen some of the fees he's paid to present at investment fund retreats and such. I know for the AI datacenter build-out, they track ~1200 datacenters world wide, to the point where they hire drone photographers to get areal shots of facilities so they can literally count how many generators or batteries or anything that are being built. By building up from the base of these numbers, you can start to project the requirements, the hardware, the supply chain, and everything else needed around compute and AI. That's what's in the spreadsheets that Semianalysis sells. 'The most expensive spreadsheets in the world'.
Like I said, the company is about 40 people now. Dylan is the head and face, but the majority of the work is done by others. We are talking people with semiconductor backgrounds, with finance backgrounds. Imagine millennials and gen Z who have grown up around technology for the most part, and a number of seasoned veterans. A number of them (including Dylan) grew up reading my content and the like. They do have stupid internal contests, such as 'who can drive the most engagement on twitter with memes on the official account', but that's meant to be the 'casual' side.
Dylan and I are quite friendly. Happy to chat and have dinner and whatnot. He's not ever asked me to join his company, and knows I'll willingly call him out on his BS, and have done regularly.
But yeah, while Dylan is Semianalysis, he doesn't do all the work. He's one person of 40. Just like Jensen is one person in ~40k.
Attached is a slide from his recent CLUSTERMAX2.0 launch, listing all the employees. Feel free do your own research.
--
This substrate thing did twang the bullshit part of my brain. There's physics, previous research, and the economics of scale. The Semianalysis article did what every other article in the ecosystem seemed to do - get overly excited at the prospect of the potential of a future promise (and because 'MERICA!
'. It meant a lot of the physics and economics discussion seemed relegated to the wayside (or put behind the paywall). It blatantly missed obvious questions and didn't do any job of trying to answer them. I'm the perennial skeptic when people come with extraordinary claims, especially ones that I can run the mathematics on. Unfortunately I think the SA article was written by someone without that mindset, which is a shame because they have been pivoting to a more STEM attitude in a lot of the coverage.I scripted and recorded a 12k word video about it, still editing. It'll be about an hour long - 30-35 minutes of it is simply going through chip design, where x-rays would work, the physics of what they do, before hitting the claims and economics. Those first 30-35 minutes will probably be super basic to anyone here, or 'no-duh' style of a high-level explainer, and the rest is basic math. Video should go up next week.
The CEO did reach out to me, about 10 minutes before my paid-for studio time was about to start. So there's a disclaimer in there that I'm due for a call. I'll likely go visit when I'm in town again for IEDM.
