OpenAI's Former Research Chief Raises $70M to Automate Manufacturing With AI

"OpenAI's former chief research officer is raising $70 million for a new startup building an AI and software platform to automate manufacturing," reports the Wall Street Journal, citing "people familiar with the matter.

"Arda, the new startup co-founded by Bob McGrew, is raising at a valuation of $700 million, according to people familiar with the matter...." Arda is developing an AI and software platform, including a video model that can analyze footage from factory floors and use it to train robots to run factories autonomously, the people said. The company's software will coordinate machines and humans across the entire production process, from product design and manufacturability to finished goods coming off the line.

The startup's goal is to make manufacturing cost effective in the Western part of the globe, reducing reliance on China as geopolitical and national security concerns rise... At OpenAI, McGrew was tasked with training robots to do tasks in the physical world, according to this LinkedIn. McGrew was also one of the earliest employees at Palantir.

Fix my ignorance

[liqu1d]

But we can't compete with China on price because of significantly lower labour costs alongside their manufacturing everything in a vertical. Unless we start ground up from mining/smelting/producing some fancy computer vision training robots is largely one of the last things we need to fix to compete again.

You can't compete with robots

[rsilvergun]

Stop worrying about competing with china. It's irrelevant and mostly just the news media trying to generate a new villain for you to get angry at so that you'll keep funding limitless defense spending now that we know Russia can't even take Ukraine.

Donald Trump's commerce Secretary admitted almost a year ago that even if the factories come back to America the jobs won't because they will be automated. The reason China hasn't done huge amounts of automation isn't cost it's because their government is int

Re:

[MpVpRb]

"we are going to have people that still need to do useful work and people that we have absolutely no profitable work"
Useful work and profitable work don't need to be coupled
We need a new economic system that allows work for the joy of doing stuff, kinda like hobbies

Yeah that's why I called it profitable work

[rsilvergun]

The problem is that the only other way to do useful work that isn't profitable is with the government and good luck doing that.

It is really easy to get people against the government doing anything for anyone but themselves. We tried to do student loan forgiveness and you literally had people who had been fucked by the student loan industry but had managed to pay off their loans demanding we don't do forgiveness because they already paid their loans.

I know queer people that get upset when other queer

Re:

[bobby]

I get what you're saying. I dunno, I'll have to think on that.

Meanwhile we have wealthy people who buy art, sponsor music / concerts, movies, etc. Down the chain those of us who have a little bit of discretionary money might buy some art, or attend a concert, or buy music, pay for a movie ticket, etc. So even in our semi-pseudo-capitalistic system, we're able to justify spending some on things that aren't life-critical necessities.

Hobbies can certainly become income-producing. If you're not so good at mark

Re:

[schwit1]

Maybe his plan is to automate existing manufacturing in the US.

You have to start somewhere. And you don't give up because one part of the supply chain is not domestic. You can import some of the parts or processed materials from overseas or get it from recycling here.

Plus, manufacturing in the US cuts down on tariffs.

Re:Fix my ignorance

[psycho12345]

Problem is... it already is heavily automated. Thus why the US can still produce the goods it does, with an ever shrinking manufacturing workforce. We have to remember, most industries work on 5 - 15 year timelines. Often the job reductions you are seeing today are the result of a process automation or improvement that was conceived of 5 years ago and has finally rolled out to the entire company or throughout an industry. It was never going to be "OH FUCK, 50% LAYOFFS!". It has been that guys that retires that isn't replaced. Or the improvement in output that leads to less hiring. See the current AI stuff in tech, it hasn't actually caused large scale layoffs, but it has stunted hiring.

Great idea!

[MpVpRb]

...in theory
In practice it's hard, really hard. Even if you believe that it's really hard, it's harder. It's not only harder than you imagine, it's harder than you can imagine.
That said, I support research on it, but don't expect success or profitability any time soon

Re:Great idea!

[serviscope_minor]

Well quite.

Many companies have trouble deploying a really basic CRUD website which has been solved technology for a few decades now. And we're supposed to expect them to have no problem deploying the still incredibly experimental tech of AI.

Likewise factories have enough trouble deploying basic PLCs (it doesn't help that anything except modbus and modbus/TCP is just awful), also decades old solved problems. But AI robots will magically be easy.

so fucking dumb

[Anonymous Coward]

The robots to make factories more efficient already exist. We know how to make factories more efficient. It just takes a buttload of capital to remake a factory. If you've ever watched How It's Made, you'll see two kinds of factories - one which looks like 1980s Mr Rogers Neighborhood where every product gets touched half a dozen times by human hands, and the other where the whole factory is basically one giant machine. We don't need some dipshit from OpenAI to tell manufacturing engineers how to do their j

Re:

[havana9]

I remember that in the late 1980s I visited a cookie factory. A the start of the line that made the butter cookies there was a "ccok" that mixed wheat, water, butter and other ingredients, then put the dough into a machine. In the middle before the packing section there was a woman doing some QA, and at the end there was a guy with a forklift loading the pallets with the cookies to the lorries.

Tangible Products People Want

[theodp]

It'd be nice to see AI used to help generate tangible products that people genuinely want and need - housing, transportation, farming, healthcare, etc. - especially if savings are passed on to consumers. Hard to get excited over the use of AI to generate task lists, emails, infinite pull requests, and "25-page reports with 100 citations" that no one wants/needs, the future of AI painted by excited execs at Microsoft's 2025 Annual Shareholders Meeting [live.com]. :-)

Invest your money in this snake oil while you can!

[thesjaakspoiler]

Imagine becoming a billionaire like Elon because you were the first to invest in this revolutionary idea.

Wrong problem

[ChrisPa]

This seems a pitch for capital rather than an idea to solve a problem.

Automated factory floors have been around a very long time. US manufacturing already uses robots extensively on the factory floor -- and in terms of output is competitive and stable absent geopolitical turmoil. There's no reason to expect an large-language model's optimization solution is going to beat existing approaches to optimization and improvement. Plus video footage seems a terrible information source for the stated goals.

And for

$70m or $700m, which is it?

[martin-boundary]

Does anybody here even read the slashdot summary anymore or just the title? Nobody has commented yet on the glaring factor of 10 discrepancy. Sloppy. Very sloppy.