ﬁSemiconductors are in virtually every electronic device that leaves a factory today, and the US government has planted a flag in the ground as the burgeoning chip industry has become the future of technology manufacturing.
Apple, Microsoft, Alphabet, Amazon, and others have been lobbying the US government to increase domestic chip production, citing problems overseas that have hampered hardware production. In fact, a US Commerce Department report in January said the chip shortage was so bad that at one point in 2021 there was just a five-day supply worldwide — with no sign the situation would improve anytime soon.
Late last month, Congress passed the CHIPS Act, a $77 billion piece of legislation that earmarks subsidies and tax credits for companies designing and manufacturing seminconductors on American shores.
Several Asian nations, including China, South Korea, and Taiwan, as well as the European Union, have also recognized the urgency in laying claim to the vast manufacturing opportunities in the semiconductor industry. They’ve injected billions in public and private funding to boost chip manufacturing capacity and development.
The tech industry in general, however, faces arguably its greatest dearth of available talent in history, with unemployment rates hovering around 1.6% — less than half that of the overall unemployment rate.
At the same time, semiconductor behemoths such as Intel, Samsung and TSMC are in the middle of building new chip fabrication plants in the US as part of a massive re-shoring effort. But there’s far too little tech talent to fill the needs.
Tony Chan Carusone
Tony Chan Curosone, a professor of electrical and computer engineering at the University of Toronto, in January became CTO at Toronto-based Alphawave IP, a five-year old, fabless semi-conductor company with offices in London and San Jose, CA. Last year, while still a flegling company developing chips for data centers, Alphawave raised $1.2 billion in an IPO led by cornerstone investors BlackRock and Janus Henderson. The IPO spoke to the massive opportunity investors see in developing application-specific integrated circuits (ASICs) for a wide variety of new technologies — from IoT devices and AI to data centers supporting cloud services.
“Computerworld” spoke with Curosone about the challenges the semiconductor faces in attracting young talent, particlarly as the industry is often seen as staid and less innovative than the software industry.
The following are experts from that interview:
Tell me a little about your background and Alphawave. “I’ve been a faculty member at the University of Toronto for 20 years, teaching electrical engineering and chip design to smart young people. Now, I’m also CTO of Alphawave. We’re focused on semi-conductors for data center connectivity. So, that’s traffic within the data center. It’s hidden from the public and it’s super high-tech stuff that’s not in your hands, but those things in your hands rely on our technology for the cloud.
“Alphawave was founded in Toronto by industry veterans. They’re geeks who founded three companies. We’ve now got 250 employees in the US and Canada. Almost all of them are engineers. And we expect to grow to over 600 [employees] by year’s end.
“So this talent recruitment thing is important to us. We’re focused on the design of the chips and engage with partners who fabricate the chips and bring them to market. So, it’s really a highly skilled workforce we’re after, entirely.
“We’re recruiting for the hardware side. I find it ironic. It should be easy. It’s the most transformative technology of our age — clearly; and yet it’s really hard. Most young, smart, ambitious people getting into the high tech think about computer engineering and software first.
“With my background in university, I’m interested in how we can increase the supply of young people and ensure there’s enough people for both hardware and software companies.”
Construction of TSMC’s 5nm semiconductor plant in Phoenix, Arizona is currently underway and scheduled to begin producing chips in 2024.
How are you dealing with the lack of talent and how your industry is different from other tech sectors in that respect? “I think part of the problem is the semiconductor industry as a whole feels like a really inaccessible technology. First, there’s a perception of semiconductors as a staid industry. It’s less exciting, there are fewer opportunities for career development and wealth creation compared to the software industry.
“You’ve got all these pictures and news stories of folks in white bunny suits doing the actual manufacturing on the factory floor. That’s one aspect that I think is a little misleading. For every person in a white bunny suit on a manufacturing floor or in a clean room, you have a whole host of engineers sitting at their desks writing code or doing all the engineering activity needed to support that manufacturing.
“The reality is semi-conductor design relies a lot on abstraction. You’re not in there engineering one molecule at a time when you’re designing a microprocessor. You’re writing code.”
“The other thing is on the technology side itself, if you’re a young, smart, ambitious person going through school and you’re interested in high tech, the semiconductor industry feels like it’s focused on a dance of molecules. The miracle there seems like it lives in quantum physics. Again, it feels less accessible to you than writing software where you can get stuff working quickly and see your immediate impact and get that feedback. That feels more exciting and dynamic. Again, though, it’s a bit of a misperception. The reality is semi-conductor design relies a lot on abstraction. You’re not in there engineering one molecule at a time when you’re designing a microprocessor. You’re writing code.
“Most of us have never set foot on a manufacturing floor. And yet, we’re designing products that are in everyone’s hands and in every data center.
“That’s what I want to emphasize. The hardware industry is one that you can hit the ground running and have an impact quickly and there’s already a lot of abstraction involved. For example, if I’m designing a chip, it’s basically writing code these days. If I showed you the code that ultimately gets turned into a chip side-by-side with code compiled for Linux, even an experienced person would have to look at the code for a few seconds to determine which is which.
“The industry is really dynamic and your day-to-day activities and the pace of things is very similar, whether you’re doing development for a software product or doing coding and design activity for a hardware product.”
Most people think of the semiconductor industry as dominated by a few dominant players. What’s the incentive for someone seeking a cutting-edge role in tech? “The general perception of the hardware industry is that R&D is slowing down. I’ve heard this comparison made: that it’s becoming like the auto industry, especially before the EV revolution when the people thought that the most innovation you might see in the auto industry is where to put the cup holder.
“But that perception comes from the idea that all it takes is one powerful CPU and you can write software on it to do whatever is needed. But the idea of the CPU was born in an age when there wasn’t enough volume required in different hardware products to justify custom designing hardware for each application. Literally, that’s what helped get Intel off the ground back in the 1960s; someone came to them asking them to build a dozen different processors for a dozen different purposes and Intel realized they could just design one chip to meet all the needs because there was not enough [sales] volume to justify doing a custom designed chip for each product.
“Now, things are changing where the volume of semiconductors is so tremendous in all these different areas. It’s become quite reasonable and desirable to custom design a solution for each type of product. So, that’s where you have different processors being designed for machine learning, whether for training or the actual online use case, or a low-power chip that’s mobile and battery operated versus one that’s used for the cloud.
“So, you have a proliferation application specific integrated circuits where just as the name suggests, you’ve got a new design for every different application. And that’s a trend that’s growing.
“With the new chip designs, that’s creating new opportunities for startups. Any startup that can identify a need for a specific application can go off and design a chip tailored for it and win that market. And it’s not a small market as, again, all ships are rising with the volume of semiconductors increasing.
“Alphawave is a good example of that. It took less than four years to go from founding with five people to unicorn IPO. That’s just one example, and there are others out there.”
What is your company doing now and how has it changed from the time it was founded?
“So we focus on connectivity solutions within the data center. We were founded in Toronto, so initially, growth was in Toronto. The founders all came from the University of Toronto. I was a faculty member at the time and still have my appointment there. Because of that strong connection, they were attracting a lot of young people right out of school. Basically, everyone was an engineer. Even today — with 250 employees — almost everyone is an engineer. So, obviously it was an environment where young people coming right out of school were just being thrown into the very cutting edge of technology. We work on the most cutting-edge fabrication technologies.
“Young people have to hit the ground running and have an impact right away. And, that was really successful — the combination of industry veterans and young people enabled us to create over 80 different product IPs in a short time. That’s really remarkable. Usually you think of hardware development as an army of people working for years to develop one product, and each person is just playing a small role. But this is a counter example where you have a small team designing 80 different products in four, four-and-a-half years.
“I think it’s a great example of what this new age of semiconductor design looks like.”
What’s the difficulty you’re running into when it comes to universities and colleges not turning out enough electrical engineers? “I think there’s a structural issue here. Young people coming into tech are drawn to the software side, and there is a ton of demand there. But the issue is universities are structurally set up so that the interest in computer software is cannibalizing electrical engineering programs.
“Universities shouldn’t be set up this way. There’s growing demand for both electrical engineers and computer engineers. So, universities have to increase capacity and not just allow computer engineering and software development programs to cannibalize electrical engineering programs.
“Then the other things that need to happen is to try to change the perception so it’s understood there’s not only a lot of opportunity for young people to hit the ground running and have an impact, but there’s also opportunity for wealth creation. Alphawave is one example of that.
“There’s a natural realignment of salary expectations happening. Semiconductor companies are increasing pay to compete. But there’s a time consequence to this too, and so that will be shaking out over the next year or two.”
What do you need most right now? “We need people that are writing code that can be turned into chips. We write code that describes a chip design and then put it through compilers that get turned into silicon. We need people writing code to validate those designs; we need people writing firmware that will ultimately run on those chips. And we need a way to make sure that firmware will run on our hardware and it’s all going to work on the application before we get the hardware back from fabrication. So, we need a whole other layer of code that’s for our verification environment.
“Those are huge teams. That’s a lot of our growth [as] we move from 250 to 600 people this year.”
Copyright © 2022 IDG Communications, Inc.
Source by www.computerworld.com