Experts agree: semiconductor manufacturing is a critical capacity that Australia must develop. But how do we start?
In the world of manufacturing, small has fast become large, with the sales of semiconductors reaching US$439 billion across the world last year. That equates to the production of nearly 300 billion semiconductor chips, or about 40 chips for every person on the planet.
Australia is no more than a bit player in this massive market, with just a few niche manufacturers of semiconductor chips, which are at the heart of all electronics, switching electrical currents on and off and controlling their direction of flow. But the country is being urged to stake a greater claim in an industry that’s now as much about strategy as profit.
“Semiconductors have strategically become the most important single production item in the world,” says Mike Nicholls, a partner at Main Sequence, Australia’s deep-tech investment fund, founded by CSIRO. “This is a $450-billion-a-year business, and Australia has virtually zero in it.”
To truly grasp the size of the semiconductor industry, ponder the almost unfathomable rate at which transistors – the semiconductor devices that are the foundation blocks of electronic devices – are being manufactured.
“We should be called the transistor-making ape, because we have made 1022 transistors, which is a trillion for every person on Earth,” says Michael Fuhrer, physics professor at Monash University and director of the ARC Centre of Excellence in Future Low-Energy Electronics Technologies. “We’ve made more transistors than we’ve made by far anything else.”
The global semiconductor manufacturing market is dominated by Taiwan, with one company – the Taiwan Semiconductor Manufacturing Company (TSMC) – producing around 55% of the world’s semiconductors. China and the US are rising manufacturers, with both countries viewing local chip production as vital to national interest. In its most recent five-year plan China called chip independence a top national priority, while four months ago the US Senate passed a bill to spend $52 billion to boost national production of semiconductors. The European Union has also declared sovereignty in semiconductor manufacture to be a priority.
A key driver behind this push for localised production has been the COVID-19 pandemic, with supply chains of vital semiconductors heavily tested – sometimes to near breaking point.
“There was a quite a shock to the industry,” says Fuhrer. “Everybody started working from home, and they wanted two things: they wanted computers to work at home and they wanted Xboxes so they can play at home.
“So there was a big demand for certain kinds of electronic products with semiconductors in them. And at the same time, a lot of industries said ‘well, demand is dropping for things like cars, so maybe we’ll scale back our orders’, but then that demand picked up again. And the supply chain is very long. So what happened is there were shortages of semiconductors for all sorts of products – General Motors had to stop manufacturing their utes, because they didn’t have the computer chips to put in them.
“If you don’t have the chips, you can’t manufacture all kinds of things. They go into almost every single industry and, of course, they’re important for defence – every kind of different piece of defensive equipment relies on advanced semiconductors and so it’s very important to have that secure supply chain.”
But if large nations are securing and controlling the production and supply of semiconductors, where does this leave a country such as Australia? Quite simply, says Nicholls, Australia needs to somehow join them in designing and manufacturing semiconductors.
“Once upon a time we had a semiconductor industry – and a relatively vibrant one in the early days, as I understand it – but we really haven’t spent much time thinking about it,” he says. “I think this recent turn of events with COVID, and everybody now realising there’s a supply-chain shortage, is actually raising awareness here in Australia.”
To create a semiconductor manufacturing plant (known as a ‘fab’ or ‘foundry’) is an investment so enormous that it can appear bamboozling and overwhelming for a country such as Australia.
“A fab is…a massive factory,” says Nicholls. “It’s probably the most complex manufacturing process of all humankind. [There are] literally hundreds and hundreds of processes to build a fab.
“If we want to set one of those up in Australia – and I think eventually, we absolutely should – we would need probably 10,000 to 15,000 highly trained semiconductor engineers to come to Australia. We need probably somewhere between $5 billion and $20 billion to set up that facility. And we’d need a bit of patience – to set up a fabrication facility for us is probably a five- to 10-year process.”
While setting up a fab requires this massive investment in money, people and time, both Nicholls and Fuhrer suggest Australia should begin at a different point along the chain – the design of semiconductors.
“The fab is massive,” says Nicholls. “The fab has big machines, lots of people, and it gets all the attention, but in fact the design process is where most of the value is created. I think we can actually compete in the design space in Australia. It may be a long time before we’re ready to build our own semiconductors here from a fabrication perspective, but I think we can absolutely compete in creating a semiconductor design industry here in Australia.
“People talk about software leading the world, but software is nothing without processes and semiconductors – every piece of software has to run on a semiconductor. Australia can have a role to play in this area, but it needs to play in the right space, and in a space that we can actually get into meaningfully in the next five years. And then maybe later, we’ll be able to jump into full fabrication. But most of the value capture happens in the design stage.
“My view is we start with designing semiconductors, and we work out how to launch 30 or 40 new start-ups in this space.”
Fuhrer suggests that the investment in design would amount to hundreds of millions of dollars rather than tens of billions, making it a more accessible and realistic option. And it could be the start of the pathway towards local manufacturing, which he believes should be targeted and strategic. Why take on the mainstream players such as TSMC and Samsung, which has a 15% share of the global semiconductor market, when Australia could potentially gain an edge by focussing on more specialised areas?
“We’re at a time where the industry is really changing,” he says. “The manufacture of advanced semiconductor chips is kind of reaching the limits of what they can do, and there’s going to be a period where that’s going to become a bit more commodified. There will be a few companies around the world that can manufacture the best devices, and they will just keep cranking those out.
“But there’s a lot of interest in some kinds of computing that are a bit different than the sort we call von Neumann architecture, general-purpose computing. For example, neuromorphic computing: making a computer chip that works a bit more like your brain, which has a lot of advantages for certain kinds of computing.
“Quantum computing is another one. Australia is a real leader in quantum computing. If we want to jump into manufacturing, that’s the kind of thing we should be looking at. What’s the next thing to come? And how can we be a leader there? I think that would be the right way to look at it strategically – how we’re going to be a leader in the decades to come.
“My recommendation for the Australian Government is that they invest in the research right now – the advanced research facilities that we think we’re going to need to make the next computer chip. Not the ones we have now, but the ones that we’re going to want in the future. That gives us the research expertise, it allows us to generate the intellectual property, and it allows us to train the workforce that’s going to be manufacturing those computer chips in the future, rather than training the workforce to manufacture the computer chips of the past.”
Nicholls says it’s vital that the investment extend to people – the semiconductor engineers who can make all of this possible. For Australia to succeed in gaining an edge in semiconductor manufacturing, we need the brainpower to create it.
“We actually don’t have a lot of semiconductor engineers in Australia,” he says. “It’s maybe in the hundreds, maybe early thousands. How do we educate our existing electrical engineers, graduates and researchers – because this is quite a different process to the general electrical engineer training that would happen at university – [about] how to actually build a processor, and all the different phases of that?
“We need to educate our local engineers, and fast track the import of fantastic semiconductor expertise from places like India, which does have a really active industry and has a great college set-up specifically for semiconductors, and also from the US and Europe as required.
Andrew Bain is a contributor to Cosmos Magazine based in Hobart, Tasmania.