If this business can’t own their IP there is no business. So why would anyone build a business that gave it all away?

Share

aesc’s Business Model

I sat down with Daniel Schultz, founder of aesc silicon, who’s pioneering a new semiconductor business model based on open source principles. Daniel is bootstrapping his startup. The model is not entirely new: his approach mirrors Linux’s (Red Hat) success. The core IP is free (for others), but the idea is that the value comes from support, customization, and specialized services. Revenue comes less from the product, more in the support.

Tim told us this (sorta), when we were discussing wafer.space’s success: similar to Linux, lots of users means lots of demand for improvement, and that ends up driving design and improvement faster than any traditional IP model could.

If the success of linux is any indication, this could work.

Who Buys Open Source?

It’s early days for Daniel and the open-source silicon movement. But a customer is important for a business to survive . . . so who’s buying?

Daniel points to verifiable security as the killer app: silicon that, because it’s open source through the entire design and manufacturing supply chain, it can be verified back-door free. Chips that have auditability (you can verify nothing was slipped into your design before it hit the fab) will have an edge in cryptographic engines must be open to be trusted.

A growing source of potential customers is the increasing demand for custom chips from smaller companies. As more companies come up with more specialized chips, this has driven a benevolent cycle for fabless companies. Open source IP could drive that faster, with less time and money locked up in IP licensing.

Daniel's bet is that falling tool costs (no Cadence/Synopsys seat) plus open IP lower the barrier so far that lots of small companies start making custom chips, and that's the demand wave. The Raspberry Pi example (one engineer made their own microcontroller) is the concrete proof point.

IP Forge

We got a sneak preview into some open source software Daniel’s building to support this vision, IP Forge. I actually found it snooping around the aesc github account before our interview.

The concept is a package manager for open-source IP blocks. Similar to NPM or python packages (shout out to uv, my favorite python package manager!) a designer could source, compare, and download blocks of code right into their project.

I like this: IP Forge seems like a crucial tool to get customers locked in to an environment like python’s pip or uv package managers, where development speed is accelerated by rapidly building on other’s building blocks.

Experimentation Gets Cheap

Daniel touched on one of the super powers of open-source and the tools that are growing up around it: the cheapness of experimentation. This could be the near-term factor that drives growth in the open-silicon space.

The $4-7k Wafer.Space runs let companies "try and error" instead of betting the R&D budget. We see that in Tim’s experiments, and in the Tiny-Tapeout runs: the price is low enough, engineers can take some true risks. Daniel points this out to me: there are open-source FPGAs being taped out, RISC-V processors going on wafer.space, and BlenderGDS is making chip renders that double as marketing only because the designs are open.

Inexpensive experimentation harnesses the power of biological evolution: change the design, experiment, try and die. Open source software further lowers the price.

The founder's wager.

The best part of the interview for me was talking to Daniel about why he chose to walk off on his own. He’s taking a very long view of the ecosystem, avoiding VC by choice, bootstrapped, a real legal entity, and he candidly says that "it's probably not the time to make a lot of money now, but open source will be a thing and it's fun."

I admire the long-range vision Daniel has for the ecosystem, and his courage for making a bet on himself and his vision.

Leave a comment

Further reading & resources

More from the Siliconimist:

• wafer.space (Tim Ansell)

• Tiny Tapeout (Matt Venn)

• InchFab (Mitchell Hsing)

• Why Fabless Ate Semis

aesc silicon and Daniel’s projects:

• aesc silicon

• Daniel Schultz on LinkedIn

• aesc silicon on GitHub

• ElemRV: end-to-end open-source RISC-V microcontroller

This is a small diversion from my usual focus on semiconductors. I live in Almaty Kazakhstan. I love the place: one of the most beautiful cities in the world, the people are some of the warmest I have ever met. From time to time, I want to share technology and finance observations I’ve seen here as an outsider.

GITEX: Making Data Centers

This week, the GITEX conference came to Almaty and I dropped in to get a better sense of what technology central Asia is focused on. I was hoping for some AI startups, but the belle of the ball this year was data centers.

Across the region, players are starting to make bets on data centers and compute centers, driven by the changing landscape between Washington and Beijing. Uzbekistan is investing in semiconductors, and data centers. Kazakhstan is investing in data centers, building a “Data Center Valley”.

For Kazakhstan, this makes sense: copious energy, open trade relations with the east and west, a man-in-the-middle location, wide open spaces - it is a great place to build data centers. (Carnegie Endowment did a great report last month on this, read here).

The hottest item was data cabling: dozens of vendors from every corner of China lined the walls with booths offering some variation of data cabling. While a few companies like Intel and Yandex had some booth space, the hustle seemed to be around racks, servers, and cabling to connect these new data centers.

Subscribe now

The Robots

The most fun part was watching the robots. I don’t think our jobs are going to be taken by them soon. But watch out if you’re planning a career in interpretive dance; see the competition below. Most of the applications seemed to be policing; surveilling your (or other people’s) citizens will be a growth industry; I predict there will be no security jobs lost to the robot dogs.

The Police Dogs

These cute little dogs could be patrolling Almaty any day now, replacing the local police. They seem too small to strap a firearm on their back, and they also seem to have some really good use cases around disaster search and rescue (Almaty is prone to earthquakes).

The Police Cars from China

Robot police cars are the most interesting robots to me, particularly, because of some of the geopolitical ramifications. This new police vehicle straight out of China, a gift (Trojan horse?) to the Kazakh police from China Aid (“For Shared Future”). China Aid is the Chinese version of USAID (not to be confused with ChinaAid.org, a non-profit protecting Chinese christians from persecution). They are (were) similar institutions: both institutions are the government’s foreign aid arms. While USAID used to focus on immunizations, and famines, and energy, China Aid is blazing a different route and is focused on . . . automated policing tools.

Kazakhstan has been fortunate enough to finance its own belts and roads so far, and hasn’t needed China’s disastrous loans. However, this police car, loaded with a small Huawei data center in the trunk, bristling with cameras and sensors, undoubtedly ties the local traffic tightly with a data center back in Beijing.

Leave a comment

Taxis

Ride sharing in Kazakhstan is dominated by the Russian search company, Yandex (Яндекс). The app will drive you and your skis across town, deliver a Dodo pizza in the depths of winter, or taxi you from Almaty to Bishkek. Many of these jobs will be done by robots soon.

Food Delivery Robots

Yandex also does food delivery in Kazakhstan. Seeing the Yandex food delivery robot, and envisioning the demise of the Yandex Food Delivery Guy made me almost as sad as the China AID surveillance car.

Yandex food delivery is an amazing Almaty fixture. These guys (all young men) are warriors, riding mopeds across the city to get you your Dodo Pizza. They drive slowly in the right lane of interstate highways, fast on the sidewalks, in the snow, on the ice, slaloming snow plows. They wear huge insulated yellow Yandex bags and they always wear a helmet.

They don’t speak English, don’t use turn signals, don’t complain, and they don’t ask for a tip. They just bang on your door, throw you your food, yell “Рақмет” and get back on their mighty moped steed with their big yellow box bags, to deliver the next pizza.

Seeing them replaced by a dopey little six wheeled Wall-E knockoff drains some of the romance of Kazakhstan. The sidewalks will be safer, but at what cost?

Yandex food delivery robots were first launched in Arizona, and were a victim of the Ukraine war. Almaty, for all its blessings in beauty, has a pretty rough sidewalk system. Even when it’s warm, dry, and snow-bank free, it’s hard to go running here and not just focus on falling into an irrigation ditch or uneven sidewalk. I wish the Yandex robots good luck with this transition, the delivery guys might have a fighting chance for now.

Peak Robot?

Freedom Financial is a local bank, a fierce competitor to Kaspi in the fintech space. They are a bank. And they’re apparently launching some dog robots. It’s not clear to me what their end goal is here (I miss a lot that’s in Russian or Kazakh) but I'm curious what they plan to do with them. It feels a little like Allbirds becoming an company.

Starlink in Kazakhstan

I've spent much of my professional life working abroad. Internet in the 2000’s and 2010’s was always the pivotal question about a place before moving there. Fueled by my fantasy of spending workdays on horseback or train, I thought it was finally my time to get Starlink; it would be a business expense as a good backup for when the internet went down.

However, due to some concerns about control, Starlink did not come to Almaty until this year. As of this year they have an authorized reseller, and they’ve moved off the map from “Coming Soon” to “Available”.

I’ve been so pleasantly surprised by how good the internet is here though, and consistently get 1g up and down in my office, and 5G in a lot of the country whenever I travel.

The only place I haven’t gotten 5G was in Mangystau last week.

There is a machine in the Netherlands that is essentially the heartbeat of the modern world. It’s called an Extreme Ultraviolet (EUV) lithography tool, and it’s built by ASML.

Without these machines, we don’t have the high-end chips that power your iPhone, your AI models, or your data centers. Each unit costs roughly $400 million, has 100,000+ parts, and requires three Boeing 747s just to deliver it.

But right now, the most coveted product coming out of ASML is the 1,000-piece Lego version.

The project started with Rick Lenssen, a data analyst at ASML, with a hobby of designing custom Lego sets, and applied it to his day job. What began as a 25,000-brick replica of the company campus eventually turned into a mission to shrink the world’s most complex tool into a box of plastic bricks.

High-Tech Scarcity

The irony of the ASML Lego set is that it mimics the scarcity of the semiconductor industry itself:

• Employee Only: You cannot buy these in stores. They are sold exclusively to ASML employees with a strictly enforced “one per person” rule.

• Secondary Market Heat: Sets have surfaced on eBay for $600, with full collections reaching as high as $4,500.

• The Sales Gap: To date, employees have bought 1,355 units of Lenssen’s latest Lego model (the EXE:5000C). In that same timeframe, ASML has sold only six of the real machines.

Why Does This Matter?

Beyond the novelty, there’s a deeper “human” layer here. Engineers often struggle to explain their hyper-complex (but meaningful) work to friends and families. Even though the world runs on it, and it involves high-powered lasers and mirrors polished to atomic-level precision . . . it’s hard to be a data analyst in an office and feel like your work is important.

The Lego set gives knowledge workers a bridge. It’s a way to take the $400 million “magic box” and turn it into something a child (or a spouse, or just about anyone) can touch, build, and understand.

Even at the absolute bleeding edge of human physics, we still have a fundamental desire to play.

We are rare.

So I took the opportunity to talk about two of my favorit things: America, and semiconductors. It’s an interesting subject, and semiconductor history ties into American culture. I like history, I’m an engineer, and I like stories . . . it’s rare I get to tie those three things together.

An American in Almaty

Being an American here feels like being a celebrity. The people, who are by far the kindest and most polite people (rivaling Omani’s who are also incredibly generous and polite), they’re genuinely curious about you and what you think of Kazakhstan. It’s not hard to find nice things to say about the place.

The lingua-franca of this region is Russian (Kazakhstan was colonized by Russia, once by the Tsars and once by the Soviets), and living here requires you to know just a bit of it. When I stepped off the plane two years ago, I did not understand a word of Russian or Kazakh, and even worse, I couldn’t read the alphabet (street signs and names can still be challenging). Kazakhs always forgive you if you don’t speak Kazakh or Russian; if they speak English, even a few phrases, they’re game to try and talk to you.

One of the most endearing things here is when you meet someone realizes you’re an American, they don’t speak English, and they grab their eldest child, and proudly tell them to speak English to you. Brave little kids come up to me all the time, at the airport, on the street, at a restaurant . . . they come right up and speak English to you. I meet so many Kazakhs that have lived and worked in the US, they always have kind things to say about my people.

So this recording was my talk, at the American Spaces, about the history of semiconductors and why they were invented in America.

Semiconductors in America

Semiconductors go all the way back to Edison; he discovered the photoelectric effect (“Edison effect” . . . a foundational idea behind solid state physics) but didn’t explore it further than naming it because he was a man in a hurry, eager to make some money.

He did this at his Menlo park labs, the first commercial lab made to make inventions, which was a model for Bell Labs.

Bell Labs, an invention factory and Nobel Prize factory, was where Shockley thought of the semiconductor, and invented it. Later leaving in a hurry, eager to make his own fortune. He was a terrible manager, managed to attract and then repel the most brilliant minds of his day, who quit and formed their own company. Not because he was an anti-democratic racist eugenicist, but because they couldn’t get anything done with him in charge. Then those brilliant minds quit their company and formed their own companies, repeating the process, dividing like cells to eventually grow into the organism known today as Silicon Valley. It’s a story of immigrants, invention, idealism, anti-trust law, and practicality that is very American, and was fun to share.

I tried to be controversial with the title “This could only have happened in America”. I did this to be provocative and try to get some reaction or pushback, but Kazakhs are too polite to push back. However, the more I thought about it, the more I realized the title was pretty accurate.

Core American values like free speech, irreverence to title and position, meritocratic operations, our openness to outsiders and outside ideas . . . these were necessary, and almost sufficient, ideas that had to be a part of the culture that created the world’s greatest engineering achievement.

This talk is my attempt to try to tie American values, the American 250th anniversary, and semiconductors together.

Happy Birthday America!

Second, Matt is the founder of Tiny Tapeout, a platform that has put nearly 3,000 chip designs into real silicon, and in the process built what might be the most accessible on-ramp to custom chip design that’s ever existed.

Together, those two things form a kind of end-to-end pipeline: learn to design, then actually build a real chip that no one else has really pulled off before.

Along the way, Matt has built a sprawling empire of STEM education focused on semiconductors, with content spanning technology, education, and the semiconductor industry.

Subscribe now

From Zero To ASIC

Custom chip design has often been viewed as an exclusive realm reserved for seasoned engineers and large corporations. However, as Matt explains, this landscape is changing rapidly; in part because of pioneering work done by Efabless and wafer.space (previous episode here). His services require no expensive tools, no NDAs and no massive non-recurring engineering (NRE) costs.

Zero to ASIC started as a course; Matt put a call up online for a course offering to help students walk through making a chip design, using the Google Open Chip Program at Skywater’s foundry. His call to action immediately recruited 300 students to the course, which has grown to use different foundries and technologies.

Over the last few years, Matt has built Zero to ASIC into an achievable path to custom chip design, developing the resources and community surrounding this field.

Tiny Tapeout

Tiny Tapeout gets open-source designers a tiny bit of real-estate on a wafer coming out of a real fab. While Matt started sending Zero to ASIC students through the Google project, he quickly switched to Efabless and ultimately wafer.space to run student designs.

With the tools they’ve created, the work to design a chip can be done right in the browser, circumventing the need to install tools. Matt’s ultimate vision: chip design becomes as straightforward and widespread as PCB prototyping, fueled by open-source innovation and collaborative effort.

chip design becomes as straightforward and widespread as PCB prototyping,

Matt emphasizes the crucial role of accessible tools, the community’s contribution to open IP libraries, and the importance of democratizing access to chip fabrication technology.

Tiny Tapeout is cheap enough you can just throw it on a credit card.

What will the world look like with 10x more chip designers? By lowering the price (and hassle) by 10x, Matt enables a world where every undergrad can tape out their own chip..

Advice to a Young Engineer

Matt’s advice to a talented, curious 22-year-old electrical engineering graduate: just do it. Get on the next tapeout!

To learn, nothing beats doing. And Tiny Tapeout enables any student to participate in designing chips from end-to-end.

Navigating the world of custom chip design may seem daunting, but with the right resources and community support, it can be an accessible and rewarding journey. Students can start with basic projects, utilize the free lessons on Matt’s website, and engage with the community . . . anyone can explore the possibilities of ASIC design.

Check out Tiny Tapeout and the Zero to ASICs course to get started.

Show Notes

We explore how Tiny Tapeout is revolutionizing chip prototyping by making it more accessible, affordable, and educational. Matt Venn shares insights into the modern landscape of chip design, the role of open source tools, and the future of semiconductor workforce development.

Key topics:

• The genesis and mission of Tiny Tapeout as a low-cost, multi-project wafer platform

• How open source tools and PDKs are making chip design accessible to amateurs and students

• The significance of multi-project wafers in reducing NRE costs and enabling first-time chip makers

• Strategies for engaging educational institutions through online and in-person workshops

• The evolution of the user base from hobbyists to industry professionals

• The importance of IP libraries and the community-driven creation of open-source chip design assets

• Challenges and opportunities in multi-foundry ecosystems and global supply chains

• The declining trajectory of Moore’s Law and the rise of custom, specialized chips

• The future vision: a Silicon community akin to PCB aggregation platforms

Timestamps:

00:00 - Introduction to Matt Venn and Tiny Tapeout’s mission
02:00 - The story behind Zero to ASICs course and open source silicon tools
04:00 - How Micro-Opportunities and Lottery programs enabled first chips
06:00 - The inception of Tiny Tapeout and its multi-project wafer approach
07:40 - Cost benefits of multi-project wafers for hobbyists and startups
09:00 - Building engagement through online and in-person workshops
10:20 - Inside the IP library and sharing designs across community members
12:00 - Challenges in community-curated IP and multi-foundry coordination
14:00 - The shifting user base from hobbyists to academia to industry
15:30 - How open tools reduce barriers for private labs and startups
16:50 - Comparing chip design to PCB workflows and simplifying the process
18:10 - Verification, costs, and the role of open source tools in ASIC development
20:00 - Supply chain resilience through multiple foundries and partnerships
22:00 - Global workforce and supply chain trends affecting semiconductor industry
24:00 - The impact of geopolitical developments on chip manufacturing
26:00 - The true limits of Moore’s Law and the future of specialized chips
28:00 - Opportunities in AI, GPUs, and custom chip design for innovation
30:00 - The importance of workforce growth and community-driven IP creation
34:00 - Expanding access via multiple foundries and reduced costs
38:00 - The long-term vision: affordable, community-driven silicon infrastructure

Resources & Links:

• Zero to ASICs Course

• Tiny Tapeout

• Gauss, Swiss Chips Partnership

Connect with Matt Venn:

• LinkedIn

“Open source is fundamentally transforming hardware.”

We talk about how lowering costs and increasing accessibility can revolutionize the industry and inspire and empower a new wave of creators.

Subscribe now

Wafer Space

wafer.space arose from the ashes of open silicon leader Efabless; when the pioneering firm shut down, Tim sprang into action, founding wafer.space to fill the void in the open silicon community.

“Cost strategies are key to democratizing hardware.”

In this episode we dive deep on the business model for open silicon, and how wafer.space is learning from history, and building back better.

Wafer space offers silicon to anyone; the platform effectively removes the "wealth barrier" to custom silicon, making it as easy to order a chip as it is to order a custom PCB.

Wafer.space’s second run has just opened up on CrowdSupply, so if you’ve always wanted to create your own semiconductor, now is your chance!

Share

Learn More

👉 wafer.space

👉 wafer.space on Github

👉 wafer.space on LinkedIn

👉 Tim Ansell on LinkedIn

👉 John Cole on LinkedIn

This is the story of how one professor led a team of students to design the first chip in Kazakhstan's history, and in doing so, may have birthed a whole new national industry.

TL:DR:

• The Win: Kazakhstan’s first student-designed RISC-V chip.

• The Strategy: Skipping the “fab-first” trap by specializing in verification via Texer.AI and Reasonbase.io

• The Takeaway: Strategy + intellectual capital > Physical capital. You don’t need a fab to own a piece of the global supply chain.

Kazakhstan’s First Chip

Nursultan’s journey began at AMD in the United States, where he worked on silicon verification. When he returned to Kazakhstan in 2023 to teach engineering, he found a desert; the local semiconductor landscape was limited mostly to FPGA programming and encryption devices. When he shared his dream of bringing "real" silicon design to the country, he was met with skepticism. In his own words, he was laughed at. Most people jumped straight to the "billion-dollar fab" problem, assuming that without a massive manufacturing plant, the industry couldn't exist.

Nursultan kept looking for an inspiring project to give his students a hands on project, a project that would move the needle through talent rather than infrastructure. Realizing that the key to developing an ecosystem here needed to start with talent, Nursultan focused in on design and verification. He found his catalyst in the “One Student, one Chip Initiative” from the Chinese Academy of Sciences. The program gave him a framework to get started, and after recruiting some enthusiastic students at Nazarbayev University they began designing the first chips.

Through a partnership with a Chinese fab, the team got space on Multi-Project Wafer (MPW) and received back their RISC-V general purpose processor. The project went viral, the media caught on, and the project gained national attention.

The project completely shifted the conversation, as Nursultan puts it,

Before people were laughing . . . but after the chip came back, people started pointing me in different directions and mechanisms . . .

From the beginning, Nursultans’ vision has looked beyond the academic track, with a plan to build something even bigger in Kazakhstan.

Bootstrapping the Ecosystem

Nursultan didn't just want a trophy for the university trophy case; he wanted a self-sustaining industry. To turn this academic momentum into a commercial reality, he and his team launched Texer.AI, a verification company designed to employ the very students who went through his program. Further along, he also co-founded ReasonBase.io, a similar company bridging the SF Bay Area and Astana, to apply verification logic to other industries.

By building a company around the chip’s success, Nursultan has created a “revolving door” between the classroom and the market. Through Texer.AI, specialized publications, and local conferences, he is guiding students directly into high-value engineering roles. This is bootstrapping at its finest: using a single successful project to fund the talent pipeline that will power the next ten projects.

The Verification Specialization

The engine driving this revolving door is a deliberate choice to skip the “fab-first” mentality. The strategy is clever: specialize in Verification. In the semiconductor world, verification is the rigorous process of mathematically proving a design works before a single dollar is spent on manufacturing.

Verification offers a low-barrier, high-margin entry point. It requires mastery of complex tools and logic rather than multi-billion dollar cleanrooms. This “capital-efficient” on-ramp allows a startup like Texer.AI to establish global credibility and attract international partnerships without needing a local factory. By focusing on high-skill employment first, Nursultan’s vision for Kazakhstan is seeding a talent pipeline that is far more essential to long-term growth than any piece of hardware.

Kickstarting a Global Blueprint

This story carries a resonance far beyond Almaty, especially as more nations chase the dream of “semiconductor sovereignty.” Nursultan’s path provides a blueprint for emerging economies: you don’t need a massive capital investment to join the global supply chain.

Innovation can start with niche expertise. By focusing on segments like design, verification, or encryption, a country can build its capabilities incrementally. Shifting the focus away from the “full manufacturing plant” requirement opens doors for any economy with a strong mathematical and engineering foundation to develop a resilient, sovereign industry.

Shifting from the assumption that you need a full manufacturing plant to participate in semiconductors opens pathways for emerging economies to develop a sovereign and resilient industry.

Nursultan’s vision for Kazakhstan, an approach of cultivating verified processor design and leveraging open-source tools exemplifies how targeted, skill-based initiatives create foundational assets for broader industry development.

Kazakhstan’s Future

Ultimately, the future of the Kazakh ecosystem relies on expanding these specialized clusters. By starting with a simple RISC-V processor and moving into complex supply chain applications, Nursultan has proven that the “impossible” is actually just a series of small, iterative steps.

Creating a core niche and nurturing the startups within it fosters a self-reinforcing cycle of talent and investment. Kazakhstan is no longer “ecosystem absent”—it is on its way to becoming a vibrant, sustainable innovation hub, one verified line of code at a time.

Share

Keeping Up With Nursultan

You can keep up with Nursultan and his sprawling semiconductor empire:

• 👉 ReasonBase on the Web

• 👉 Nursultan Kabylkas on LinkedIn

• 👉 Nursultan on the Web

• 👉 John Cole on LinkedIn

The next semiconductor revolution won’t just sit in a server rack. It will be embedded in drywall, aircraft wings, and bridges.

Imagine adding instrumentation to infrastructure, airplanes, or even your office with a roll of tape that senses, computes, and updates in real time. The underlying devices are fabricated using traditional semiconductor processes. They’re then integrated into a flexible substrate, networked, and deployed directly onto physical surfaces to gather data at scale.

An Aha Moment

Co-founder Mike Chieco reveals how a simple idea, inspired by his three-year-old with tape, evolved into a platform that could reshape industries from data centers to aerospace. Complex concepts like Physical AI and high-density sensing suddenly became tangible. Instead of pitching abstractions, Mike and Mike could toss a roll of tape across the table and let the product explain itself.

University Tech Transfer

FlyingWire is built on technology developed in a university lab by Mike Filler at Georgia Tech. Mike Chieco and I dive into the licensing process: what it takes to commercialize university IP, how he identified the core value, and where the tech transfer system works, and where it doesn’t.

The Bigger Trend: Semiconductors Moving Into Materials

For fifty years, semiconductor progress was defined by shrinking transistors. Moore’s Law, lithography advances, nodes. . . . these were the axis of innovation. Value was concentrated inside the die.

That model is maturing. The next frontier is not smaller logic gates; it’s where silicon lives. Semiconductors are migrating out of boxes and circuit boards and into substrates, surfaces, and structures themselves. Rather than mount hardware onto infrastructure, sensing, compute, and connectivity are being integrated directly into materials: tape, composites, aircraft skin, building surfaces. The real innovation is less about node size and more about packaging.

This shift changes the economics and the industry. Historically, installation labor often exceeded the cost of the silicon itself; embedding semiconductors into materials collapses that friction and unlocks infrastructure-scale instrumentation. It elevates advanced packaging and substrate engineering to strategic importance and opens entirely new TAM categories in aerospace, energy, defense, and data infrastructure. The next semiconductor battleground won’t just be leading-edge fabs, it will be the integration of silicon into the physical world, where chips don’t just power machines, but become part of them.

FlyingWire is compelling precisely because it sits at the center of that transition.

More on FlyingWire

Mike writes about tech transfer, the origins of FlyingWire, and semiconductors more broadly on Substack.

Dispatches From the Frontier

The Second Frontier: Flying Wire

For New Readers: Welcome. These dispatches started as my documentation of searching for the next company to build. That search is over. Now I’m building…

Read more

7 months ago · 3 likes · Mike Chieco

Connect with the founders, and read more below:

• Mike Chieco on LinkedIn

• Mike Filler on LinkedIn

• John Cole on LinkedIn

The Siliconimist

The Siliconimist has candid conversations with semiconductor enthusiasts, about chips, capital, and the geopolitical consequences. We talk to entrepreneurs, investors, technologists and operators that are pushing semiconductors to new heights.

By John Cole

You too can own a fab in your backyard, for just a few million USD . . . we’ll talk about why you should get one.

• InchFab on the Web

• InchFab on LinkedIn

• Mitchell Hsing

• John Cole

Understanding Microfabrication

Microfabrication is an intricate process that allows for the manufacturing of devices and materials at a microscopic scale. Inch Fab is pioneering this field by providing application-specific fabs that drastically reduce the cost and footprint required for traditional fabrication methods. According to Mitchell, their entire fab line occupies just about 100 square meters and costs between $5 to $10 million. This is a significant reduction compared to the billions often required for larger fabs.

The Origin of Inch Fab

In a world of mega fabs, gigafabs, and terrafabs, why’d Mitchell go small? The company's inception traces back to the PhD research conducted by Mitchell and his co-founder at MIT, where the objective was to democratize the fabrication process. They aimed to lower the barriers to entry, enabling more innovators to create at the micro scale.

As Mitchell recalls, their early equipment was often outdated, sometimes requiring the use of floppy disks, showcasing how innovation can thrive even with limited resources.

A Shift in Business Model

We discuss how Inch Fab has shifted its business model from providing foundry services to offering manufacturing lines. This strategic pivot responds to a growing global demand for decentralized fabrication solutions. As industries increasingly seek flexibility and cost-effectiveness, Inch Fab’s model proves timely and necessary.

Democratizing Fabrication

Inch Fab's mission is not purely educational; rather, it aims to empower individuals and smaller companies that may not have the resources to engage with larger semiconductor fabs. Mitchell emphasizes that approximately 80% of current products do not justify the economics of a multi-billion dollar fab, indicating a vast market potential for smaller-scale fabrication. This approach opens the door for innovation across various sectors that previously struggled with high entry costs.

The Future of Microfabrication

Finally, we take some time to discuss emerging trends in semiconductor manufacturing, including the shift towards application-specific designs. The barriers of traditional fabrication are being challenged as new material stacks and innovative designs take center stage. Mitchell notes that the future will likely see more diverse applications and the rise of technologies such as quantum computing and diamond-based electronics.

In this episode:

• How an American executive ended up managing semiconductor fabs in China

• The evolution of China’s semiconductor goals from self-sufficiency to supply chain independence

• The influence of Big Fund initiatives on China’s industry growth

• The effects of US sanctions on China’s technological development and acquisition strategies

• China’s ongoing push for advanced nodes and alternative lithography methods

• The potential for a stable US-China semiconductor balance and future innovation pathways

Resources & Links:

• A Decade in the Chinese Semiconductor Industry, An American Story (Amazon Affiliate Link)

• Connect with Doug LinkedIn

Why are they showing up twice? They play two distinct roles, and appear as:

• Fabless system designers. At the top of the ecosystem, they define workloads, select architectures, and set specs.

• OEMs/platform owners. At the bottom of the ecosystem (or the consumer end of the pipe) they ship and operate the end system.

That dual role changes everything. They’re still customers, but they’re no longer just customers. In short: they don’t just buy chips - they shape what gets built.

How Did We Get Here?

Why did hyperscalers moved into chip design. The “fabless takeover” story is driven by four forces in the industry:

1) Workload-specific performance beats general-purpose

General compute load has gotten heavier, and specialty chips have become increasingly viable. AI training/inference, recommender systems, video, search: the gains come from tailoring compute + memory + interconnect to their workloads. This drives economics, better performance per Watt, per dollar, and total utilization (the hidden killer metric in data centers).

2) The bottleneck is system-level, not chip-level

Once you’re at scale with these chips, and have scaled demand, the limiting factors become memory bandwidth, networking, power & cooling. Owning the silicon roadmap lets hyperscalers optimize the whole stack.

3) Supply chain power & allocation

Custom silicon is also a procurement weapon. If you’re running design in house, it slows down IP leakage. But it can also help secure better terms for purchase, and reduce dependency on IDM roadmaps.

4) Economic capture

If you operate at hyperscale, and you’re looking for a place to deploy capital, you don’t just want margin. You want rent capture via platform control (cloud services, hardware/software co-optimization).

Where is this heading?

Fabless Is Becoming System-Dominant—and Foundries Are Becoming Subordinate.

The center of gravity in semiconductors is shifting away from manufacturing leadership toward system ownership. Hyperscalers will increasingly define the workloads that matter - along with memory, interconnects, and packaging architectures.

In that world, foundries behave less like technology leaders and more like capacity providers, where how fast they deliver may matter more than abstract node leadership. Power accrues to those who control demand at scale, not necessarily to those with the most advanced process.

As a result, foundries compete on reliability, geopolitics, and allocation as much as on nanometers. This dynamic helps explain why companies like Apple, Google, and Amazon can punch far above their weight without owning fabs - and why future semiconductor winners may look more like platform companies than traditional chip manufacturers.

What ASML is actually doing is re-optimizing for speed. Specifically, engineering speed. The company is trimming leadership and management layers in Tech and IT while explicitly adding engineering capacity and reassigning many former managers back into individual contributor roles. That distinction matters.

This is not a retrenchment. It’s a structural signal.

A rare, high-quality talent release valve

For startups in hardware, industrial software, equipment, and deeptech, this creates a short window of opportunity. ASML doesn’t employ generic managers. It employs people who have:

• shipped systems that take years to assemble,

• coordinated multi-physics engineering across suppliers,

• lived inside safety-critical, capital-intensive production environments.

Even if many of the displaced roles skew toward leadership, those individuals tend to have deep technical backgrounds. Some were engineers who were pushed into management as the company scaled. This is one of the cleanest talent inflection points European deeptech has seen in years.

The quiet opportunity: engineering-throughput startups

ASML’s internal language is unusually candid: engineers are spending too much time coordinating, reporting, and managing slow process flows. For startups, this is the most interesting signal of all.

There is growing demand for products that:

• reduce coordination cost in complex hardware programs,

• automate verification, validation, and traceability,

• manage long-lived codebases and configuration drift,

• close the loop between design, test, and manufacturing data.

Not generic SaaS. Not dashboards for executives. Tools that directly increase engineering throughput in environments where mistakes cost millions.

ASML’s move is a reminder that even the most elite industrial organizations are still fighting friction at the seams.

What startup founders do next?

If you’re building in or around the semiconductor stack, these are your opportunities:

• Hiring: Look for ex-ASML talent who owned outcomes, not headcount.

• Product: Sell cycle time, validation speed, and decision quality, not “productivity.”

• Positioning: Frame your company as removing friction from engineering, not adding another layer.

ASML is choosing to fix organizational drag while it’s still winning. That should tell startups two things: first, speed still matters, even at the top of the stack. Second, companies that help engineers move faster will stay relevant no matter where the cycle turns.

If you, dear entrepreneur, are building for that world, the signal is very strong.

Coming to you live from Almaty, Kazakhstan . . .

This is straight out of The Innovator’s Dilemma.

I recently re-watched a lecture by Clay Christensen (author of Innovators Dilemma and a few other masterpieces.

Clay Christensen’s used the steel mini-mills Japan used to take over the steel market to illustrate how incumbents retreat to high-margin activities when new entrants emerge. In his classic steel mini-mills example, Clay Christensen showed how incumbents don’t lose because they’re dumb; they lose because they’re rational. When low-margin entrants attack the bottom of the market, incumbents flee to higher-margin segments. Each move looks smart on a quarterly basis. Collectively, it’s fatal.

As Christen points out, US Steel managers weren’t stupid: Every time a low margin mini-mill came into a market, they were motivated to flee to higher margin markets. If you’re David (mini-mills), you’ve created a situation where Goliath (US Steel) is motivated to run away from you rather than fight you. This is especially true if you’re timeline is quarterly: as a manager, you’re making the smart move away from a highly competitive market.

Today, we see the incumbent DRAM manufacturers starting to retreat to high-margin DRAM products while emergents (CXMT) start with low-margin products and shove high-margin makers out of the market.

But if current memory prices are driven by OpenAi shenanigans, when the bottom falls out of the AI market, which DRAM maker is going to survive? My guess: the state-supported trailing-technology provider will survive, while SK Hynix, Samsung, and Micron become the US-Steels of yesteryear. Still large, but strategically hollowed out.

Free Link to WSJ Article covering CXMT.

This podcast was made with AI, and is only holding place here until I get some episodes with real humans up!