semiconductor

Stop Supplying. Start Owning.

Canada has a paradox that I have been talking about for a long time.

We are home to most of the world’s AI godfathers. We have Nobel Prize winners, world-class researchers, and some of the most respected engineering schools on the planet. Those institutions attract exceptional students from around the world. And in turn, Canada trains some of the best engineering talent anywhere.

Yet an increasing number of those graduates — now approaching 80% — leave after convocation.

With them go the startups they might have built, the economic value they might have created, and the wealth that could have stayed here. We all know the largest and most valuable companies are technology companies, and they are based in the United States. Many of them were co-founded or led by Canadians.

It is a story of could have been, would have been, and should have been.

I was recently invited to speak at the Deans of Engineering Conference in Winnipeg. I want to share the core of what I said, because I think this conversation matters well beyond that room.

The root cause is not what most people think

The easy explanations are talent, capital, and policy. We hear them constantly. They are not wrong, but they are not the root cause.

The root cause is a mindset.

Clearly, we do not have a shortage of ambition or ability. One fundamental issue stands out — yet very few people talk about it, let alone address it. We have collectively learned to think like suppliers, not owners.

Without the owner mindset, we unintentionally and subconsciously optimize for producing great talent for other countries, rather than building a stake in where that talent goes and what it creates.

This is the supplier mindset made visible.

Let me illustrate with a banana

I use a banana analogy in my talks because it makes the concept concrete.

If you put money and a banana in front of a monkey, the monkey takes the banana. It does not know that money can buy many bananas. That insight comes from Jack Ma, who used this analogy to compare the mindset of someone who grabs a job versus someone who builds a company.

Jobs are bananas. They are real, they matter, and they feed people. A banana feeds you once. Cash feeds you many times.

But the lessons go further.

A banana tree is better than a banana because it can feed you forever. Ownership compounds; wages do not. A banana farm is better than a tree — because participation as a supplier is not enough; ownership of the platform is the real prize. And a store is better than a farm, because the store owns the customer relationship and captures the value that flows through the entire chain.

This is the progression from employee thinking to owner thinking. From banana to store. From grabbing to owning.

Canada has been grabbing bananas

Let me make this concrete with two examples — one personal, one national.

The Wattpad calculation. When Wattpad was acquired for US$660 million, the headline was a Canadian success story. And in many ways, it was. But here is the number nobody talks about. By the time of the acquisition, roughly half of the company was owned by Canadians. When the deal closed, about US$330 million in economic value left the country — because we had raised capital from outside Canada to build it.

Wattpad’s annual payroll was roughly US$30 million. Not small. But compared to the acquisition price, it is a fraction. Ownership creates far more value than employment. Jobs matter. Entrepreneurship matters. But nothing compares to owning world-class companies.

The auto industry analogy. Many people say Canada has a strong auto industry. We do not. We have a strong auto supplier industry. That is not the same thing. Our auto suppliers — collectively — are worth a fraction of GM, Ford, or Toyota. They build the factories, employ the workers, and take on the operational risk. When the EV transition stalled, the suppliers’ brand new facilities went quiet. When the majors slowed production, the layoffs rippled through.

The supplier bears the downside. The owner captures the upside and sets the rules.

And when the Canadian government went to attract EV investment, what did we do? We signed deals to become suppliers again — subsidized by Canadian taxpayers, while the ownership, brand, and margin stayed elsewhere. We took the risk but not the profit.

This is the supplier mindset at a national scale.

The question nobody asks clearly enough

Here is the hinge question: what does winning actually look like?

The supplier mindset and the owner mindset do not just lead to different outcomes. They lead to completely different definitions of winning.

If you are a supplier, sending your best researchers to OpenAI is a win. You produced world-class talent. Mission accomplished. That belongs in the annual report.

If you are an owner, that is a loss. You invested in that person for years, and you ended up owning nothing. The outcome looks identical from the outside — a brilliant Canadian thriving on the world stage — but the two mindsets score it completely differently.

Until we agree on what winning actually means, we will keep celebrating losses as victories.

Where the mindset gets formed

Here is what I have come to believe: the supplier mindset is not learned on the job. It is learned in school.

The mental model a student builds about what success looks like — a FAANG job offer, a US grad school acceptance, a signing bonus from a company they can brag about — is set before they ever enter the workforce.

In the US, building a unicorn startup is Plan A. Getting a job at Google is Plan B. In Canada, getting a job at Google — or going to the US — is Plan A. Building a startup, let alone a unicorn, is often not even in the equation.

That quote is from a world-class Canadian AI scientist who is now at a US company. It landed hard when I first heard it, because it is accurate.

The deans are the front line

I said something direct to the room in Winnipeg that I want to say here too.

The founders of most of Canada’s future tech giants are sitting in engineering classrooms right now. The deans who lead those schools are the single most underleveraged force in Canada’s innovation economy.

Here is why this is also in the deans’ own interest. If their students build world-class companies and keep them here, those companies will forever be associated with that school. That is a legacy that compounds for decades. The next crop of students is inspired by the tech giants that exist. Right now, leading universities outside of Canada are winning that recruitment battle — not because their engineering programs are better, but because the companies their graduates built are more visible, more celebrated, and more aspirational.

Think about OpenAI. It was co-founded by a University of Toronto alumnus. Most people associate it with Silicon Valley.

That association is not fixed. It is a choice, made one graduating class at a time.

This is not only about encouraging entrepreneurship

I want to be precise here, because there is a version of this argument that deans hear all the time and that I think misses the point.

Many engineering schools already encourage entrepreneurship. Hackathons. Incubators. Pitch competitions. These are necessary. But they do not define what success looks like. And in a strange way, encouraging entrepreneurship is still a supplier mindset — we are producing entrepreneurs for the ecosystem and hoping something sticks.

The real call to action is different. It is to start and scale world-class companies here in Canada.

That is a higher bar. A different ambition. A fundamentally different culture to build. It means celebrating the founder who builds a billion-dollar Canadian company with the same institutional pride as the researcher who wins a Nobel Prize. It means changing what the school defines as a win — not only placements, publications, and patents, but also companies that stay, scale, and own their category.

The window is now

I have saved the most important point for last.

We lost Game 1. Canada invented modern AI. The most important AI companies are almost all based in the US. That window has closed.

But Game 2 is underway. Quantum computing. Robotics. Physical AI. Space. Advanced manufacturing. Smart energy. Just to name a few. Canada has deep roots in all of these — world-class labs, exceptional researchers, and early-stage companies that are genuinely competitive.

Here is what is different about Game 2: you cannot pack up a quantum computing facility or a robotics lab and move it to San Francisco. Unlike software, the physical infrastructure is sticky. The talent clusters around it. The companies that emerge will be rooted where the labs are.

And the ground-level signal I am seeing is genuinely encouraging. I have never met more professors and researchers who want to start companies — and who want to do it in Canada. That is new. That is meaningful.

The conditions are finally aligned to address the root problem, not just the symptoms. But the only trophy that ultimately matters is homegrown, world-class companies. And we can only win Game 2 — and ultimately the championship — if we build the owner mindset now, starting with the people who shape how the next generation of engineers think about what success looks like.

Addressing the supplier mindset and turning it into an owner mindset can create the domino effect that turns Canada’s bragging rights into lasting economic wins.

That is the game we can win.

Portfolio Highlight: Zinite. Speed and Energy, Two Birds, One Stone

For most of semiconductor history, progress was a simple loop. Shrink transistors. Fit more into the same area. Get faster compute as a byproduct.

That loop had a name. Moore’s Law. It traces back to Intel co-founder Gordon Moore. He observed in the 1960s that the number of transistors on a chip, and hence its capabilities, tended to double every two years. The industry turned that observation into a roadmap. It was never guaranteed to run forever. Now shrinking is harder because we are starting to hit many limits in physics and economics, and the cost of pushing the frontier keeps rising.

So if the curve is going to keep bending upward, the industry needs new scaling vectors beyond making everything smaller in two dimensions.

This is why Two Small Fish invested in Zinite in 2021 at the company’s inception. The thesis was simple then, and it is still simple now. Scale in the third dimension, using proprietary technology protected by patents to enable true 3D chips.

Zinite stayed deliberately stealth early on, focused on building the core and protecting it properly before saying too much. Five years after we invested, we can finally talk about it more openly.

The company is led by its CEO, Dr. Gem Shoute. Fun fact. Her breakthrough was strong enough that her professors and industry veterans (who helped create fundamental IP used in all chips since 2008) joined her as co-founders, Dr. Doug Barlage and Dr. Ken Cadien.

The Distance Tax

In a recent blog post, I used a factory analogy to explain why speed, latency, and energy are often bottlenecked by movement, not necessarily arithmetic.

In short, systems don’t lose because they can’t do math. GPUs are already very good at that. Systems lose speed because they can’t feed the math with data fast enough.

In many systems, moving data costs far more than doing the arithmetic. When movement is expensive, speed and energy efficiency get worse together.

AI inference exacerbates the problem because the computational characteristics of AI inference workloads put a premium on memory behaviour. In many cases, the limiting factor is not arithmetic. It is how efficiently the system can move data. Bringing memory closer to logic matters because it directly reduces that movement.

Sensing fits in the same frame as logic and memory. Sensors generate raw data at high volume. If the system’s first step is to ship raw data far away before anything useful happens, it pays in bandwidth, latency, and power. The more intelligence that can happen closer to where data is produced, the less the system wastes just transporting information.

So the distance tax is one big problem showing up in three places at once. Logic. Memory. Sensing.

Why 3D Matters for Speed and Energy

When people hear 3D chips, they think density. More transistors per area. That matters. The bigger lever is proximity. Current 3D approaches to deliver more performance per area rely on advanced packaging, which is hindered by cost and the distance tax.

If memory can live closer to logic, the system avoids transfers that dominate both performance and power. If compute and memory can sit closer to sensing, the system avoids hauling raw streams around before doing anything intelligent.

Every avoided transfer is a double win. Speed improves because stalls go down and effective bandwidth goes up. Energy improves because fewer joules are burned moving bits instead of doing work.

That is the two birds, one stone result.

Five years after we invested, Zinite is far from just a concept. The company is doing exceptionally well, and it represents the kind of platform that can extend performance gains into the post-Moore era by reducing the distance tax, not by asking physics for more shrink, but by making data travel less.

Portfolio Highlight: ABR’s Funding Round

Edge AI has been a key pillar of our Advanced Computing Hardware investments and a core part of our thesis for a long time. It is the same arc I wrote about in The Next Data Centre: Your Phone a while ago.

We need new architectures to meet the speed, security, and energy demands of the next frontier of computing and its applications, which is the lens I used in The Factory Analogy.

Our portfolio company Applied Brain Research (ABR) just achieved a new milestone: ABR announced the successful closure of its oversubscribed seed funding round, including investment from TSF as a lead investor, with Eva Lau joining the board.

ABR created and patented a new type of AI model, called state space models, to make AI smaller, faster, and more energy efficient than transformer models. State space models deliver real-time voice and time series intelligence without the cloud, built for privacy and efficiency. ABR’s first chip, TSP1, delivers real-time, fully on-device voice AI without the cloud. Full vocabulary speech-to-text and text-to-speech are now possible at under 30mW.

At the edge, every millisecond and every milliwatt count.

For context:

30mW is 100× less than a 3W LED lightbulb.
A data-center GPU lives in a different universe: an NVIDIA H200 NVL is up to 600W.

Now connect that to the three constraints that define the edge:

Speed: for voice and interaction, half a second is half a second too late. Cloud voice is “a terrible experience,” plagued by delays.
Security: shipping voice data to the cloud bakes in privacy risk by default — which is why we keep coming back to intelligence that stays close to the user, as Brandon argued in his post In Favour of Intelligence That Stays Put. ABR calls out “privacy concerns” as a core issue with cloud voice.
Energy: edge devices are constrained by battery life and on-device resources. ABR’s on-device voice numbers move this from “interesting” to “deployable.”

This is why ABR enables numerous new use cases that weren’t viable before in categories like AR, robotics, wearables, medical devices, and automotive.

Imagine AR glasses (or other wearables) that respond to your command in real time without draining the battery. Imagine a robot that reacts with no hesitation. Imagine a medical device that can provide insight securely, without exporting sensitive data. Imagine a car that can respond to voice commands even when the network is unreliable. These are just a few examples. The list can go on and on.

Or as Eva put it in ABR’s announcement: sophisticated voice AI doesn’t require the cloud.

Five Areas Shaping the Next Frontier

The cost of intelligence is dropping at an unprecedented rate. Just as the drop in the cost of computing unlocked the PC era and the drop in the cost of connectivity enabled the internet era, falling costs today are driving explosive demand for AI adoption. That demand creates opportunity on the supply side too, in the infrastructure, energy, and technologies needed to support and scale this shift.

In our Thesis 3.0, we highlighted how this AI-driven platform shift will reshape behaviour at massive scale. But identifying the how also means knowing where to look.

Every era of technology has a set of areas where breakthroughs cluster, where infrastructure, capital, and talent converge to create the conditions for outsized returns. For the age of intelligent systems, we see five such areas, each distinct but deeply interconnected.

1. Vertical AI Platforms

After large language models, the next wave of value creation will come from Vertical AI Platforms that combine proprietary data, hard-to-replicate models, and orchestration layers designed for complex and large-scale needs.

Built on unique datasets, workflows, and algorithms that are difficult to imitate, these platforms create proprietary intelligence layers that are increasingly agentic. They can actively make decisions, initiate actions, and shape workflows. This makes them both defensible and transformative, even when part of the foundation rests on commodity models.

This shift from passive tools to active participants marks a profound change in how entire sectors operate.

2. Physical AI

The past two decades of digital transformation mostly played out behind screens. The next era brings AI into the physical world.

Physical AI spans autonomous devices, robotics, and AI-powered equipment that can perceive, act, and adapt in real environments. From warehouse automation to industrial robotics to autonomous mobility, this is where algorithms leave the lab and step into society.

We are still early in this curve. Just as industrial machinery transformed factories in the nineteenth century, Physical AI will reshape industries that rely on labour-intensive, precision-demanding, or hazardous work.

The companies that succeed will combine world-class AI models with robust hardware integration and build the trust that humans place in systems operating alongside them every day.

3. AI Infrastructure

Every transformative technology wave has required new infrastructure that is robust, reliable, and efficient. For AI, this means going beyond raw compute to ensure systems that are secure, safe, and trustworthy at scale.

We need security, safety, efficiency, and trustworthiness as first-class priorities. That means building the tools, frameworks, and protocols that make AI more energy efficient, explainable, and interoperable.

The infrastructure layer determines not only who can build AI, but who can trust it. And trust is ultimately what drives adoption.

4. Advanced Computing Hardware

Every computing revolution has been powered by a revolution in hardware. Just as the transistor enabled mainframes and the microprocessor ushered in personal computing, the next era will be defined by breakthroughs in semiconductors and specialized architectures.

From custom chips to new communication fabrics, hardware is what makes new classes of AI and computation possible, both in the cloud and on the edge. But it is not only about raw compute power. The winners will also tackle energy efficiency, latency, and connectivity, areas that become bottlenecks as models scale.

As Moore’s Law hits its limit, we are entering an age of architectural innovation with neuromorphic computing, photonics, quantum computing, and other advances. Much like the steam engine once unlocked new industries, these architectures will redefine what is computationally possible. This is deep tech meeting industrial adoption, and those who can scale it will capture immense value.

5. Smart Energy

Every technological leap has demanded a new energy paradigm. The electrification era was powered by the grid. Today, AI and computing are demanding unprecedented amounts of energy, and the grid as it exists cannot sustain this future.

This is why smart energy is not peripheral, but central. From new energy sources to intelligent distribution networks, the way we generate, store, and allocate energy is being reimagined. The idea of programmable energy, where supply and demand adapt dynamically using AI, will become as fundamental to the AI era as packet switching was to the internet.

Here, deep engineering meets societal need. Without resilient and efficient energy, AI progress stalls. With it, the future scales.

Shaping What Comes Next

The drop in the cost of intelligence is driving demand at a scale we have never seen before. That demand creates opportunity on the supply side too, in the platforms, hardware, energy, physical systems, and infrastructure that make this future possible.

The five areas — Vertical AI Platforms, Physical AI, AI Infrastructure, Advanced Computing Hardware, and Smart Energy — represent the biggest opportunities of this era. They are not isolated. They form an interconnected landscape where advances in one accelerate breakthroughs in the others.

We are domain experts in these five areas. The TSF team brings technical, product and commercialization expertise that helps founders build and scale in precisely these spaces. We are uniquely qualified to do so.

At Two Small Fish, this is the canvas for the next generation of 100x companies. We are excited to partner with the founders building in these areas globally, those who not only see the future, but are already shaping it.

P.S. If you enjoyed this blog post, please take a minute to like, comment, subscribe and share. Thank you for reading!

This blog is licensed under a Creative Commons Attribution 4.0 International License. You are free to copy, redistribute, remix, transform, and build upon the material for any purpose, even commercially, as long as appropriate credit is given.

Legends of Semiconductors: The Only Dinner Where the Edition Number Matters

At most dinners, introductions start with your name and maybe what you do.

At this one, we began with:
“Second edition.”
“Fourth edition.”

Why? Because this was our “School of Fish – Legends of Semiconductors” dinner, hosted at our home, where your relationship with the Sedra & Smith textbook was the common thread.
(I’m second edition, if you’re wondering.)

We were incredibly honoured to have Dr. Adel Sedra, former Dean of Engineering at the University of Waterloo, join us. Recently appointed to the Order of Canada, Dr. Sedra is a towering figure in the world of electrical engineering. Since 1982, his textbook has taught more than three-quarters of the world’s electrical engineers. It is hard to find someone in the field who has not studied from it. I consider myself extraordinarily fortunate, not just to have learned from his book, but to have been his student more than 30 years ago at the University of Toronto. Few have had the privilege of learning directly from a legend.

We were equally honoured to host Benny Lau, co-founder of ATI Technologies, whose legacy lives on in AMD’s GPUs to this day. AMD acquired ATI for $5.4 billion nearly 20 years ago, still one of the largest tech acquisitions in Canadian history. When Eva worked at ATI, she had the chance to work closely with Benny. His presence brought our conversation full circle, from classroom to commercialization. Adding even more depth to the evening, Benny was also once a student of Dr. Sedra. Two generations of engineers at the same table, both shaped by the same teacher.

From left to right: Benny Lau, Eva Lau, Ljubisa Bajic

This evening was also a chance to reconnect with those who shaped my own journey. Martin Snelgrove and Raymond Chik, my professor and TA respectively, were both there and are now serial entrepreneurs. They are also co-founders of Hepzibah, a Two Small Fish portfolio company. (I still can’t help but sometimes call him Professor Snelgrove.) Xerxes Wania, another one of my TAs from back in the day, went on to build and exit two semiconductor companies and added his voice to the conversation.

From left to right: Xerxes Wania, Dr. Adel Sedra, Allen Lau, Martin Snelgrove, Raymond Chik

We were also joined by Ljubisa Bajic, former CEO of TensTorrent and now CEO of Taalas, who also spent part of his career at ATI, further adding to the thread that connected many of us. Chris Yip, Dean of Engineering at the University of Toronto, and Deepa Kundur, current Chair of U of T’s Department of Electrical & Computer Engineering—continuing the legacy of leadership that Dr. Sedra once held in that position—also attended. Professor Tony Chan Carusone, now also CTO of Alphawave Semi and coauthor of the Sedra & Smith textbook starting with the 8th edition, brought both academic and commercial perspectives to the table.

From the TSF portfolio side, we were thrilled to have Professor Doug Barlage of the University of Alberta and Professor Chris Eliasmith of the University of Waterloo, co-founders of Zinite and ABR, respectively.

And of course, our partner Dr. Albert Chen joined us. He is a graduate of Waterloo Engineering and knows a thing or two about semiconductors himself.

Semiconductors brought us together that night.
Textbook and tapeout were what we talked about, and we all loved them.

P.S. If you enjoyed this blog post, please take a minute to like, comment, subscribe and share. Thank you for reading!

Announcing Our Investment in Hepzibah AI

The Two Small Fish team is thrilled to announce our investment in Hepzibah AI, a new venture founded by Untether AI’s co-founders, serial entrepreneurs Martin Snelgrove and Raymond Chik, along with David Lynch and Taneem Ahmed. Their mission is to bring next-generation, energy-efficient AI inference technologies to market, transforming how AI compute is integrated into everything from consumer electronics to industrial systems. We are proud to be the lead investor in this round, and I will be joining as a board observer to support Hepzibah AI as they build the future of AI inference.

The Vision Behind Hepzibah AI

Hepzibah AI is built on the breakthrough energy-efficient AI inference compute architecture pioneered at Untether AI—but takes it even further. In addition to pushing performance/power harder, it can handle training loads like distillation, and it provides supercomputer-style networking on-chip. Their business model focuses on providing IP and core designs that chipmakers can incorporate into their system-on-chip designs. Rather than manufacturing AI chips themselves, Hepzibah AI will license its advanced AI inference IP for integration into a wide variety of devices and products.

Hepzibah AI’s tagline, “Extreme Full-stack AI: from models to metals,” perfectly encapsulates their vision. They are tackling AI from the highest levels of software optimization down to the most fundamental aspects of hardware architecture, ensuring that AI inference is not only more powerful but also dramatically more efficient.

Why does this matter? AI is rapidly becoming as indispensable as the CPU has been for the past few decades. Today, many modern chips, especially system-on-chip (SoC) devices, include a CPU or MCU core, and increasingly, those same chips will require AI capabilities to keep up with the growing demand for smarter, more efficient processing.

This approach allows Hepzibah AI to focus on programmability and adaptable hardware configurations, ensuring they stay ahead of the rapidly evolving AI landscape. By providing best-in-class AI inference IP, Hepzibah AI is in a prime position to capture this massive opportunity.

An Exceptional Founding Team

Martin Snelgrove and Raymond Chik are luminaries in this space—I’ve known them for decades. David Lynch and Taneem Ahmed also bring deep industry expertise, having spent years building and commercializing cutting-edge silicon and software products.

Their collective experience in this rapidly expanding, soon-to-be ubiquitous industry makes investing in Hepzibah AI a clear choice. We can’t wait to see what they accomplish next.

P.S. You may notice that the logo is a curled skunk. I’d like to highlight that the skunk’s eyes are zeros from the MNIST dataset. 🙂

Portfolio Highlight: ABR

The next frontier of AI lies at the edge — where data is generated. By moving AI toward the edge, we unlock real-time, efficient, and privacy-focused processing, opening the door to a wave of new opportunities. One of our most recent investments, Applied Brain Research (ABR), is leading this revolution by bringing “cloud-level” AI capabilities to edge devices.

Why is this important? Billions of power-constrained devices require substantial AI processing. Many of these devices operate offline (e.g., drones, medical devices, and industrial equipment), have access only to unreliable, slow, or high-latency networks (e.g., wearables and smart glasses), or must process data streams in real time (e.g., autonomous vehicles). Due to insufficient on-device capability, the only solution today is to send data to the cloud — a suboptimal or outright infeasible approach.

How does ABR solve this? ABR’s groundbreaking technology addresses these challenges by delivering “cloud-sized” high-performance AI on compact, ultra-low-power devices. This shift is transforming industries such as consumer electronics, healthcare, automotive, and a range of industrial applications, where latency, reliability, energy efficiency, and localized intelligence are essential.

What is ABR’s secret sauce? ABR’s unique approach is rooted in computational neuroscience. Co-founded by Dr. Chris Eliasmith, CTO and Head of the University of Waterloo’s Computational Neuroscience Research Group, ABR leverages a brain-inspired invention called the Legendre Memory Unit (LMU), which was invented by Dr. Eliasmith and his team of researchers. LMUs are provably optimal for compressing time-series data—like voice, video, sensor data, and bio-signals—enabling significant reductions in memory usage. Running the

LMU on ABR’s unique processor architecture has created a breakthrough that “kills three birds with one stone” by:

1. Increasing performance,

2. Reducing power consumption by up to 200x, and

3. Cutting costs by 10x.

This is further turbocharged by ABR’s AI toolchain, which enables customers to deploy solutions in weeks instead of months. Time is money, and ABR’s technology allows for advanced on-device functions—like natural language processing—without relying on the cloud. This unlocks entirely new use cases and possibilities.

At the helm of ABR is Kevin Conley, the CEO and a former CTO of SanDisk, alongside Dr. Chris Eliasmith. Together, they bring exceptionally strong leadership across both hardware and software domains—a rare but powerful combination that gives ABR a significant competitive advantage.

ABR’s vision aligns perfectly with our investment thesis and our belief that edge computing and software-hardware convergence represent the next frontier of opportunity in computing. We’re excited to see ABR power billions of devices in the years to come.

P.S. This blog is licensed under a Creative Commons Attribution 4.0 International License. You are free to copy, redistribute, remix, transform, and build upon the material for any purpose, even commercially, as long as appropriate credit is given.

Celebrating a Legendary Educator

I was fortunate to not only learn from his textbook but also to be a student in his class. Few have the privilege of learning directly from a legend, and I consider myself incredibly lucky to have been in the right place at the right time—more than 30 years ago—to benefit from his lectures.

Who am I talking about? Professor Adel Sedra.

I wanted to take a moment to congratulate Professor Sedra on the recognition of his incredible legacy with the launch of a new permanent exhibit at the University of Toronto. His textbook, Microelectronic Circuits, co-authored with the late Professor Kenneth C. Smith, has been a cornerstone of engineering education for decades. To date, it has gone through eight editions (with Professor Tony Chan Carusone also part of the editorial team), sold more than a million copies, and been translated into nearly a dozen languages.

Here’s a fact I only recently discovered: it’s estimated that over three-quarters of electrical engineers in the world since 1982 have studied this book—yes, 75%!—widely known as “Sedra/Smith” after its authors.

“When they first sat down in 1982 to create the first draft, I don’t think either of the two co-authors fully realized that it would become the gold standard in the field,” said Christopher Yip, Dean of U of T Engineering.

As a professor, Professor Sedra was simply unparalleled in the field of microelectronics. His passion for teaching was evident, and his exams? They were tough—though I like to think I did alright! 😉

Watching this video gave me goosebumps.

As a 20-year-old at the time, I didn’t fully comprehend or appreciate that I was sitting in a classroom with a legendary professor, studying one of the earlier editions of what would become a truly iconic textbook.

Professor Sedra’s contributions to engineering education and his impact on generations of students are unmatched. This exhibit is a fitting tribute to a man who shaped how the world learns about microelectronics.

You can read more about this celebration of his legacy here: U of T Engineering News.

Fabless + ventureLAB is Cloud Computing for Semiconductors

This is a follow-up blog post to my last piece about Blumind.

More than two decades ago, before I started my first company, I was involved with an internet startup. Back then, the internet was still in its infancy, and most companies had to host their own servers. The upfront costs were daunting—our startup’s first major purchase was hundreds of thousands of dollars in Sun Microsystems boxes that sat in our office. This significant investment was essential for operations but created a massive barrier to entry for startups.

Fast forward to 2006 when we started Wattpad. We initially used a shared hosting service that cost just $5 per month. This shift was game-changing, enabling us to bootstrap for several years before raising any capital. We also didn’t have to worry about maintaining the machines. It dramatically lowered the barrier to entry, democratizing access to the resources needed to build a tech startup because the upfront cost of starting a software company was virtually zero.

Eventually, as we scaled, we moved to AWS, which was more scalable and reliable. Apparently, we were AWS’s first customer in Canada at the time! It became more expensive as our traffic grew, but we still didn’t have to worry about maintaining our own server farm. This significantly simplified our operations.

A similar evolution has been happening in the semiconductor industry for more than two decades, thanks to the fabless model. Fabless chip manufacturing allows companies—large or small—to design their semiconductors while outsourcing fabrication to specialized foundries. Startups like Blumind leverage this model, focusing solely on designing groundbreaking technology and scaling production when necessary.

But fabrication is not the only capital-intensive aspect. There is also the need for other equipment once the chips are manufactured.

During my recent visit to ventureLAB, where Blumind is based, I saw firsthand how these startups utilize shared resources for this additional equipment. Not only is Blumind fabless, but they can also access various hardware equipment at ventureLAB without the heavy capital expenditure of owning it.

Let’s see how the chip performs at -40C!

The common perception that semiconductor startups are inherently capital-intensive couldn’t be more wrong. The fabless model—in conjunction with organizations like ventureLAB—functions much like cloud computing does for software startups, enabling semiconductor companies to build and grow with minimal upfront investment. For the most part, all they need initially are engineers’ computers to create their designs until they reach a scale that requires owning their own equipment.

Fabless chip design combined with shared resources at facilities like ventureLAB is democratizing the semiconductor space, lowering the barriers to innovation, and empowering startups to make significant advancements without the financial burden of owning fabrication facilities. Labour costs aside, the upfront cost of starting a semiconductor company like Blumind could be virtually zero too.

That’s why the saying, “software once ate the world alone; now, software and hardware consume the universe together,” is becoming true at an accelerated pace. We have already made several investments based on this theme, and we are super excited about the opportunities ahead.

Portfolio Highlight: Blumind

When it comes to watches, my go-to is a Fitbit. It may not be the most common choice, but I value practicality, especially when not having to recharge daily is a necessity to me. My Fitbit lasts about 4 to 5 days—decent, but still not perfect.

Now, imagine if we could extend that battery life to a month or even a year. The freedom and convenience would be incredible. Considering the immense computing demands of modern smartwatches, this might sound far-fetched. But that’s where our portfolio company, Blumind, comes into play.

Blumind’s ultra-low power, always-on, real-time, offline AI chip holds the potential to redefine how we think about battery life and device efficiency. This advancement enables edge computing with extended battery life, potentially lasting years – not a typo – instead of days. Products powered by Blumind can transform user behaviours and empower businesses and individuals to unlock new and impactful value (see our thesis).

Blumind’s secret lies in its brain-inspired, all-analog chip design. The human brain is renowned for its energy-efficient computing abilities. Unlike most modern chips that rely on digital systems and require continuous digital-to-analog and analog-to-digital conversions (which drain power), Blumind’s approach emulates the brain’s seamless analog processing. This unique architecture makes it perfect for power-sensitive AI applications, resulting in chips that could be up to 1000 times more energy-efficient than conventional chips, making them ideal for edge computing.

Blumind’s breakthrough technology has practical and wide-ranging applications. Here are just a few use cases:

• Always-on Keyword Detection: Integrates into various devices for continuous voice activation without excessive power usage.

• Rapid Image Recognition: Supports always-on visual wake word detection for applications such as access control, enhancing human-device interaction with real-time responses.

• Time-Series Data Processing: Processes data streams with exceptional speed for real-time analysis in areas like predictive maintenance, health monitoring, and weather forecasting.

These capabilities unlock new possibilities across multiple industries, including wearables, smart home technology, security, agriculture, medical, smart mobility, and even military and aerospace.

A few weeks ago, I visited Blumind’s team at their ventureLAB office and got an up-close look at their BM110 chip, now in its third tapeout. Blumind exemplifies the future of semiconductor startups through its fabless model, which significantly lowers the initial infrastructure costs associated with traditional semiconductor companies. With resources like ventureLAB supporting them, Blumind has managed to innovate with remarkable efficiency and sustainability. (I’ll share more about the fabless model in an upcoming post.)

I’m thrilled to see where Blumind’s journey leads and how its groundbreaking technology will transform daily life and reshape multiple industries. When devices can go years without needing a recharge instead of mere hours, that’s nothing short of game-changing.

Image: Close-up view of BM110. It is a piece of art!

Image: Qualification in action. Note that BM110 (lower-left corner) is tiny and space-efficient.

Image: The Blumind team is working hard at their ventureLAB office. More on this in a separate blog post here.

Our portfolio company, Blumind, is revolutionizing device efficiency with its ultra-low power, always-on, real-time, offline AI chip. Inspired by the human brain’s energy-efficient computing, Blumind’s innovative all-analog design significantly reduces power consumption, making its chips up to 1000 times more efficient than conventional digital chips.

This advancement enables edge computing with extended battery life, potentially lasting YEARS - not a typo - instead of days. Practical applications of Blumind’s technology include always-on keyword detection for voice activation, rapid image recognition for access control, and real-time time-series data analysis for predictive maintenance and health monitoring. These capabilities unlock new and previously impossible opportunities across various industries, from wearables and smart homes to security, agriculture, military, and aerospace.

Recently, I visited Blumind’s team at their ventureLAB office and witnessed their third-tapeout BM110 chip in action. I’m excited to see Blumind’s continued growth and how its transformative technology will reshape industries, making long-lasting, energy-efficient devices a reality.

The Next Data Centre: Your Phone

The history of computing has been a constant shift of the centre of gravity.

When mainframe computers were invented in the middle of the last century, they were housed in air-conditioned, room-sized metal boxes that occupied thousands of square feet. People accessed these computers through dumb terminals, which were more like black and white screens and keyboards hooked to the computer through long cables. They were called dumb terminals because the smart part was all on the mainframes.

These computers worked in silos. Computer networks were very primitive. Data was mainly transferred through (physical!) punch cards and tapes.

The business model was selling hardware. During that era, giants like IBM and Wang emerged, and many subsequently submerged.

Hardware was the champion.

Mainframe computers in the 50s. Image source: Wikipedia

The PC era, which started in the 80s and supercharged in the 90s, ended the reign of the mainframe era. As computers became much faster while the price dropped by orders of magnitude, access to computing became democratized, and computers appeared on every desktop. We wanted these computers to talk to each other. Punch cards clearly no longer worked as there were millions of computers now. As a result, LANs (local area networks) were popularized by companies like Novell, which enabled the client/server architecture. Unlike the previous era, the “brains” were decentralized, with clients doing much of the heavy lifting. Servers still played a role, but for the most part, it was for centralized storage.

Although IBM invented the PCs, the business models shifted, creating the duopoly of Intel (and by association companies like Compaq) and Microsoft, with the latter capturing even more value than the former. The software era had begun.

Software became the champion. Hardware was dethroned to the runner-up.

Then, in the late 90s to the 2010s, the (broadband) web, mobile, and cloud computing came along. Connectivity became much less of an issue. Clients, especially your phones, continued to improve at a fast pace, but the capability of servers increased even faster. The “brains” shifted back to the server as that’s where the data is centralized. For the most part, clients were now responsible for user experience, important but merely a means to an end (of collecting data) rather than an end in themselves.

Initially, it appeared that the software-hardware duopoly would continue as companies like Netscape and Cisco were red hot, only to be dethroned by companies like Yahoo and AOL and later Google and Meta. Software and hardware were still crucial, but they became the enablers as the business model once again shifted.

Data became the newly crowned champion.

Fast forward to now, the latest—and arguably the greatest of all time—platform shift, powered by generative AI, is upon us. The ground beneath us is shifting again. On a per-user basis, generative AI demands orders of magnitude more energy. At a time when data centres are already consuming more energy than many countries, it is set to double again in two years to roughly equivalent to the electricity consumption of Japan. The lean startup era is gone. AI startups need to raise much more capital upfront than previous generations of startups because of the enormous cost of compute.

Expecting the server in the data centres to do all the heavy lifting can’t be sustainable in the long term for many reasons. The “brains” have once again started to shift back to the clients at the edge, and it is already happening. For instance, Tesla’s self-driving decisions are not going to make the round trip to its servers. Otherwise, the latency will make the split-second decisions a second too late. Another example, most people may not realize this, but Apple is an edge computing company already as its chips have had AI capabilities for years. Imagine how much more developers can do on your iPhone—at no cost to them—instead of paying a cloud provider to run some AI. That would be the Napster moment for AI companies!

Inevitably, now that almost every device can run some AI and is connected, things will be more decentralized.

In past eras, computing architectures evolved due to the constraints of—or the liberation of—computing capabilities, connectivity, or power consumption. The landscape has once again shifted. Like past platform shifts, there will be a new world order. The playing field will be levelled. Rules will be rewritten. Business models will be reinvented. Most excitingly, new giants will be created.

Every. Single. Time.

Seeing the future is our superpower. That’s why a while ago, at Two Small Fish Ventures, we have already revised our thesis. Now, it is all about investing in the next frontier of computing and its applications, with edge computing an important part of it. Our recent investments have been all-in on this thesis. If you are a founder of an early-stage, rule-rewriting company that is taking advantage of this massive platform shift, don’t hesitate to reach out to us. We love backing category creators in massive market opportunities.

We are all engineers, product builders and company creators. We know how things work. Let’s build the next champion together!

Update: This blog post was published just before Apple announced Apple Intelligence. I knew nothing about Apple Intelligence at that time. It was purely a coincidence. However, it did validate what I said.

The depressing numbers of the venture-capital slump don’t tell the full story

Thank you to The Globe for publishing my second op-ed in as many weeks: The depressing numbers of the venture-capital slump don’t tell the full story.

The piece is now available in full here:

Bright spots in the current venture capital landscape exist. You just need to know where to look.

Recent reports are right. Amid high interest rates, venture capitalists have a shrinking pool of cash to dole out to hopeful startups, making it more challenging for those companies to raise funding. In the United States, for example, startup investors handed out US$ 170.6 billion in 2023, a decrease of nearly 30 percent from the year before.

But the headline numbers don’t tell the whole story.

There’s a night-and-day difference between the experience of raising funds for game-changing, deep-technology startups that specialize in artificial intelligence and related fields, such as semiconductors, and those who try to innovate with what’s referred to as shallow tech.

Remember the late 2000s? Apple’s App Store wasn’t groundbreaking in terms of technical innovation, but it nonetheless deserves praise because it revolutionized the smartphone. Then, the App Store’s charts were dominated by simplistic applications from infamous fart apps to iBeer, the app that let you pretend you were drinking from your iPhone.

That’s the difference – those building game-changing tools and those whose products are simply trying to ride the wave.

Tons of startups are pitching themselves as AI or deep-tech companies, but few actually are. This is why many are having trouble raising funds in the current climate.

It’s also why the era of shallow tech is over, and why deep-tech innovations will reshape our world from here on out.

Toronto-based Ideogram, a deep-tech startup, was the first in the industry to integrate text and typography into AI-generated images. (Disclosure: This is a company that is part of my Two Small Fish Ventures portfolio. But I’m not mentioning it just because I have a stake in it. The company’s track record speaks for itself.)

Barely one year old, the startup has fostered a community of more than seven million creators who have generated more than 600 million images. It went on to close a substantial US$80-million Series A funding round.

As a comparison, Wattpad, the company I founded, which later sold for US$660-million, had raised roughly US$120-million in total. Wattpad’s Series A in 2011, five years since inception, was US$3.5-million.

The speed at which Ideogram achieved so much in such a short period of time is eye-popping.

The “platform shifts” over recent decades have largely played out in the same way. From the personal-computer revolution in the late 20th century to the widespread adoption of the internet and cloud computing in the 2000s, and then the mobile era in the 2010s, there’s a clear pattern.

Each shift unleashed a wave of innovation to create new opportunities and fundamentally reshape user behaviour, democratize access and unlock tremendous value. These shifts benefited the billions of internet users and related businesses, but they also paved the way for “shallow tech.”

The late 2000s marked the beginning of a trend where ease of creation and user experience overshadowed the depth of innovation.

When Instagram launched, it was a straightforward photo-sharing app with just a few attractive filters. Over time, driven by the massive amounts of data it collected, it evolved into one of the leading social media platforms.

This time is different. The AI platform shift makes it harder for simplistic, shallow-tech startups to succeed. Gone are the days of building a minimally viable product, accumulating vast data and then establishing a defensible market position.

We’re entering the golden age of deep-tech innovation, and in order to be successful, startups have to embrace the latest platform shift – AI. And this doesn’t happen by tacking on “AI” to a startup’s name the way many companies did with the “mobile-first” rebrand of the 2010s.

In this new era, technological depth is not just a competitive advantage but also a fundamental pillar for building successful companies that have the potential to redefine our world.

For example, OpenAI and Canada’s very own Cohere are truly game-changing AI companies that have far more technical depth than startups from the previous generation. They’ve received massive funding partly because the development of these kinds of products is very capital-intensive but also because their game-changing approach will revolutionize how we live, work and play.

Companies like these are the bright spots in an otherwise gloomy venture-capital landscape.

Software Once Ate the World Alone; Now, Software and Hardware Consume the Universe Together

Over a decade ago, in his blog post titled “Why Software is Eating the World,” Marc Andreessen explained why software was transforming industries across the globe. Software would no longer be confined to the tech sector but permeated every aspect of our lives, disrupting traditional businesses and creating new opportunities, driving innovation and reshaping the competitive landscape. Overall, the post underscores the profound impact of software on the economy and society at large.

While the prediction in his blog post was mostly accurate, today, the world is still only partially eaten up by software. Although there are many opportunities for software alone to completely transform user behaviour, upend workflow, or cause other disruptions, the low-hanging fruits are mostly picked. That’s why I said the days of shallow tech are behind us now.

Moving forward, increasingly, there will be more and more opportunities that require hardware and software to be designed and developed together from the get-go to ensure that they can work harmoniously and make an impact that otherwise would not be possible. The best example that people can relate to today is Tesla. For those who have driven a Tesla, I trust many would testify that their software and hardware work really well together. Yes, their self-driving software might be buggy. Yes, the build quality of its hardware might not be the best. However, with many features on their cars – from charging to navigation to even warming up the car remotely – you can just tell that they are not shoehorning their software and their app into their hardware or vice versa.

On the other hand, on many cars from other manufacturers, you can tell their software and hardware teams are separated by the Grand Canyon and perhaps only seriously talk to each other weeks before the car is launched 🙂

We also witness the same thing down to the silicon level. From building the next AI chip to the industrial AI revolution to space tech, software and hardware convergence is happening everywhere. For instance, the high energy required by LLMs is partially because the software “works around” the hardware, which was not designed with AI in mind in the first place. Changes are already underway, ensuring that software and hardware dance together. There is a reason why large tech players like OpenAI and Google are planning to make their own chips.

We are in the midst of a once-in-a-decade “platform shift” because of generative AI. In the last platform shift more than a decade ago, when the confluence of mobile and cloud computing created a massive disruption, there was one “iPhone moment,” and then things progressed continuously. This time, new foundation models are launching at a break-neck pace, which is further exacerbated by open-source. So fast that we are now experiencing one iPhone moment every few weeks.

All of this happens when AI-native startups are an order of magnitude more capital-intensive than in the past cycle. At the same time, investors are also willing to write big cheques to these companies, but perhaps it is appropriate, given all the massive opportunities ahead of us.

Investing in this environment is both exciting and challenging as assessing these new opportunities is drastically different from the previous-generation software-only, shallow-tech startup.

The next few years are going to be wild.

Allen's Thoughts…

Recovering CEO (Wattpad), 3x entrepreneur, deep tech VC (TSF), electrical engineer, helping founders build world-class companies