Startup Aims to Build Hundreds of Chip Factories with Prefab Parts and AI (fastcompany.com) 28
"To meet the world's growing hunger for chips, a startup wants to upend the costly semiconductor fabrication plant with a nimbler, cheaper idea..." reports Fast Company, "an AI-enabled chip factory that can be assembled and expanded modularly with prefab pieces, like high-tech Lego bricks."
In other words, they want to enable what is literally a fast company... "We're democratizing the ownership of semiconductor fabs," says Matthew Putman, referring to chip fabrication plants. Putman is the founder and CEO of Nanotronics, a New York City-based industrial AI company that deploys advanced optical solutions for detecting defects in manufacturing procedures. Its new system, called Cubefabs, combines its modular inspection tools and other equipment with AI, allowing the proposed chip factories to monitor themselves and adapt accordingly — part of what Putman calls an "autonomous factory." The bulk of the facility can be preassembled, flat-packed and put in shipping containers so that the facilities can be built "in 80% of the world," says Putman.
Eventually, the company envisions hundreds of the flower-shaped fabs around the world, starting with a prototype in New York or Kuwait that it hopes to start building by the end of the year... Nanotronics says a single Cubefab installation could start at one acre with a single fab, and grow to a four-fab, six-acre footprint. Each fab could be built in under a year, the company says, with a four-fab installation estimated to cost under $100 million. Nanotronics declined to disclose how much it has raised for the project, but Putman says the company has previously raised $170 million from investors, including Peter Thiel and Jann Tallin, the Skype cofounder...
A single automated Cubefab will need only about 30 people to operate, "and they don't have to be semiconductor experts," says Putman. "AI takes away that need for that specialization that you would normally need in a fab." [...] Putman also hopes automation will help further reduce the environmental impact of an industry that's notoriously resource-intensive and produces thousands of tons of waste a year, much of it hazardous. "Because you have the AI fixing the material and the device before it's manufactured, you have less waste of the final material," he says.
Thanks to Slashdot reader tedlistens for sharing the news.
In other words, they want to enable what is literally a fast company... "We're democratizing the ownership of semiconductor fabs," says Matthew Putman, referring to chip fabrication plants. Putman is the founder and CEO of Nanotronics, a New York City-based industrial AI company that deploys advanced optical solutions for detecting defects in manufacturing procedures. Its new system, called Cubefabs, combines its modular inspection tools and other equipment with AI, allowing the proposed chip factories to monitor themselves and adapt accordingly — part of what Putman calls an "autonomous factory." The bulk of the facility can be preassembled, flat-packed and put in shipping containers so that the facilities can be built "in 80% of the world," says Putman.
Eventually, the company envisions hundreds of the flower-shaped fabs around the world, starting with a prototype in New York or Kuwait that it hopes to start building by the end of the year... Nanotronics says a single Cubefab installation could start at one acre with a single fab, and grow to a four-fab, six-acre footprint. Each fab could be built in under a year, the company says, with a four-fab installation estimated to cost under $100 million. Nanotronics declined to disclose how much it has raised for the project, but Putman says the company has previously raised $170 million from investors, including Peter Thiel and Jann Tallin, the Skype cofounder...
A single automated Cubefab will need only about 30 people to operate, "and they don't have to be semiconductor experts," says Putman. "AI takes away that need for that specialization that you would normally need in a fab." [...] Putman also hopes automation will help further reduce the environmental impact of an industry that's notoriously resource-intensive and produces thousands of tons of waste a year, much of it hazardous. "Because you have the AI fixing the material and the device before it's manufactured, you have less waste of the final material," he says.
Thanks to Slashdot reader tedlistens for sharing the news.
Skynet Meets IKEA: Assembly Required (Score:3)
Building the factory is not the main challenge (Score:3)
Re: (Score:2)
Re: (Score:2)
Also, chemicals. He's theorizing there will be less waste as the AI overlord improves the process, but the chemicals are still bad.
As the step size is reduced, the need for chemicals goes down. A 5 nm chip has half the area of a 7 nm chip and uses half the chemicals.
Re: (Score:2)
Fast company (Score:4, Insightful)
...pulling a fast one.
Re: (Score:2)
"an AI-enabled chip factory that can be assembled and expanded modularly with prefab pieces, like high-tech Lego bricks."
They forgot to mention that it'll be blockchain-powered and funded via NFTs sold over DeFi networks. Also, is it web scale?
Re: (Score:2)
"They forgot to mention that it'll be blockchain-powered and funded via NFTs sold over DeFi networks. Also, is it web scale?"
You forgot 3D-printed Virtual Reality AI.
I have an alternative (Score:2)
You build all these production lines near each other, so that when they do need servicing you can efficiently use your pool of technicians.
I call it a fab.
Use the magic word.. (Score:3)
AI
Get funding
And a bridge to go with it too (Score:3)
A chip fab needs lots and lots of heavy expensive equipment, like air and water handlers, flywheel UPSs, toxic chemical handling equipment, just to name the obvious ones that come to mind.
All of these things are big, expensive, and have few counterparts in the downmarket space, meaning they're mostly custom installs for each fab rather than an amalgamation of stuff you can find at home depot or best buy.
What this means is that it almost always pays to amortize that fixed cost over a large factory. Meaning you can build a microfab thing but it won't compete against the big boys at all. Probably not by an order of magnitude.
This guy is selling snakeoil.
Re: And a bridge to go with it too (Score:2)
Yes, this was my first reaction. The fab itself is only part of the process. The utilities and waste handling are much more complex. Even if you build it in some country that lets you dump it into a river, you still need reliable power and lots of clean water.
Cute (Score:3)
Startup want your money... (Score:3)
... to make off with it. The sad thing is that some people will "invest" into this when it is abundantly clear this is a scam. I cannot believe anybody that has even only looked a tiny bit into actual semiconductor making can push something like this and be honest.
Re: (Score:2)
No, they don't want to make off with YOUR money. They're just need your money so they can string this out long enough to get that sweet, sweet buyout money.
Re: (Score:2)
Nobody will buy them. Unless having a demonstrated product has become optional for that now?
Brilliant! (Score:2)
Building fabs is hard, but with AI it's easy! Now we just need to apply AI to fusion and quantum computing, and those challenges will become easy, too!
There are indeed tasks where AI is making a contribution, so the current AI hype is justified or at least understandable in some cases, but there are also snake oil salespeople trying to make an quick buck.
Clearly a fraud (Score:4, Insightful)
Even if they purposely mis-use the word "fab" to mean a single stage in the whole process, it still is silly. There is no way $100 million buys enough gear to do all the steps needed to make a chip - no matter how "mature" or old. By the time you buy A SINGLE piece of each machine needed, you are into billions. If you any decent capacity, many billions.
Hell, I will bet that $100 million won't even pay for the plumbing needed for the ultra-pure water, not to mention all the nasty chemicals needed in different stages. (Fabs are the single industry that uses the most different chemicals that are really nasty.)
Re: (Score:3)
These fabs won't have EUV machines or anything else cutting edge. Read the article, it's very clear about this. These are small scale fabs for companies that want to make custom chips on older processes. Their capabilities will be similar to what you find today in many universities, just more automated. A relevant quote from the article:
VerWey estimates there are currently about 150 to 200 of these nanofabrication facilities in the U.S., most of them within university research labs. Facilities capable of producing more boutique chips for commercial uses could have positive effects for supply chain security, and could help more quickly prototype devices, helping bridge a gap between basic and applied research and real-world applications. "People have been enthused by the prospect of figuring out how to make it affordable to run these sorts of nanofabs for a long time," he said.
Re: (Score:2)
Do you know how many steps are involved in make a chip, however old or obsolete?
Re: (Score:2)
Yes, I do know something about it. I also know that your "no matter how mature or old" claim is simply false. In the semiconductor industry this is known as "Rock's law": the cost of building a fab doubles every four years. There are a few reasons for that. Smaller features need much more sophisticated equipment. Wafers have gotten larger, which makes all the equipment more expensive. And fabs have just gotten larger.
Back in the 1980s, $100 million was about the standard price for a fab. There's been
In the 1950s (Score:3)
Reminds me of predictions in the 1950s: In 10 years, computers will accurately predict the weather. In 20 years, we'll have walking, talking robots.
Computers still can't parse a limited model of the real world (3D vison-system), can't perform error-correction and compound error-correction (fall-over, then can't recover) and we're only just approaching the portable giga-flops required to estimate the answer.
AI AI AI big data!!?1one (Score:2)
I'm holding out... (Score:2)
...for the desktop fab. It's tautologically 30 years closer now than it was 30 years ago.
Not economically feasible (Score:2)
The issue of fabs, as they may eventually discover if they get the chance to run out of money, is needing highly-specialized, extremely low tolerance, well-maintained, difficult-to-configure equipment in an absurdly clean environment. Economies-of-scale make it viable, whereas aspirational approaches like these are doomed to fail like high volume 3D printing that is better served by injection molding.
Also, the emphasis of AI seems VC buzzword bingo rather than a legitimate automation process engineering co
I'll beat them to it (Score:2)
P.S. Still looking for ways to do it on the blockchain. Suggestions needed.