Can AI Help Fix Boring Software?

[00:00:00] OCI is the single platform for your infrastructure, database, application development, and AI needs.

[00:00:06] Do more and spend less like Uber, 8x8, and Databricks Mosaic.

[00:00:11] Take a free test drive of OCI at oracle.com slash wallstreet.

[00:00:19] Welcome to Tech News Briefing. It's Monday, December 2nd. I'm Julie Chang for The Wall Street Journal.

[00:00:25] Over a decade ago, venture capitalist Mark Andreessen said that software is eating the world.

[00:00:33] And today, companies like Figma are playing a role in that.

[00:00:37] Figma helps designers and product developers collaborate to build their own software products.

[00:00:43] At WSJ Tech Live in October, our global tech editor, Jason Dean, sat down with Dylan Fields, co-founder and CEO of Figma,

[00:00:51] to discuss what role artificial intelligence plays in this collaborative process.

[00:00:56] Later on in the show, we'll hear highlights from their conversation.

[00:01:00] But first, the latest episode of our new podcast series, Bold Names, dropped on Saturday right here in the Tech News Briefing feed.

[00:01:12] WSJ columnists and co-hosts of Bold Names Tim Higgins and Christopher Mims spoke with J.B. Straubel,

[00:01:19] the founder and CEO of battery recycling startup Redwood Materials,

[00:01:24] where he's on the cutting edge of an energy revolution that could turn battery recyclers into the new big oil.

[00:01:30] Before starting Redwood, Straubel was Tesla's chief technology officer, and he currently sits on the company's board of directors.

[00:01:37] In this episode, he shares what he learned at the Elon Musk School of Management.

[00:01:42] WSJ's Danny Lewis spoke with Tim Higgins and Christopher Mims about their conversation with Straubel.

[00:01:48] On your latest episode, you talked to J.B. Straubel, who spent 15 years designing batteries for electric vehicles at Tesla.

[00:01:54] But now, instead of building batteries at his new startup, Redwood Materials, Straubel's trying to break them down.

[00:02:00] Why?

[00:02:01] To save the environment, to live in the earth and have a kind of a virtuous cycle where instead of mining these materials,

[00:02:09] which can be very messy to make batteries, essentially he's mining trash to bring in an electric car future.

[00:02:16] That's the ultimate goal.

[00:02:18] It'll take him years and years to get there, but he's already showing kind of some remarkable early success just after a few years,

[00:02:24] generating revenue from the sales of being able to pull things like nickel and lithium out of used batteries

[00:02:32] and sell those raw materials back into the battery supply chain.

[00:02:37] Right. And actually, at one point in the conversation, Straubel did make a comparison to his company that might be pretty surprising for the battery business.

[00:02:45] It feels a bit like we are inventing the next generation of refineries, so to speak.

[00:02:51] You know, I hope we don't quite have a direct comparison to an ExxonMobil in the future,

[00:02:55] but, you know, there is some commonality that, you know, we are taking these old raw materials

[00:03:02] and inventing ways to refine them into a whole selection of new products that the world needs.

[00:03:07] Christopher, how could battery recyclers become the new big oil?

[00:03:11] If you look at the scale of the energy transition we're going through,

[00:03:16] EVs are maybe 2% of the fleet in the U.S. right now.

[00:03:20] So imagine a future in which EVs are 50%, right?

[00:03:25] Literally 25 times more EVs on the road than we have now or more.

[00:03:30] To sustain that, to make sure that all those vehicles have batteries,

[00:03:35] we are going to have to be recycling absolutely massive amounts of all of these materials that are inside their batteries.

[00:03:43] We just don't have a choice.

[00:03:44] You can't pull that much out of the ground.

[00:03:46] We can't build enough mines no matter how cavalier we get about the resulting environmental destruction.

[00:03:52] So the same way that we have a country full of fracking operations and a Gulf Coast full of giant oil refineries,

[00:04:02] we're going to have giant lithium and nickel and rare earth element recycling facilities in the U.S.

[00:04:08] just to reprocess that material so we can keep this electrified economy going.

[00:04:14] It's going to be EVs.

[00:04:16] It's going to be batteries in our homes.

[00:04:17] It's going to be an equal amount of batteries probably on the grid.

[00:04:22] So that kind of transformation, just the scale of it, it's almost mind-boggling.

[00:04:27] Straubel used to work at Tesla, and you talked about his relationship with Tesla owner Elon Musk.

[00:04:32] Straubel's now on the company's board of directors,

[00:04:34] and he told you one thing he takes away from how Musk runs his companies is an aggressive approach to business.

[00:04:40] I think humanity is probably underperforming a little bit versus its potential.

[00:04:45] You know, maybe Elon is kind of beyond the optimum at times,

[00:04:48] but I think, you know, it could be probably a positive example to really crank up the intensity for a lot of other folks.

[00:04:56] What else did he have to say about his experience working with Musk?

[00:04:59] I asked him if he was sleeping on the factory floor, because you have to remember,

[00:05:03] Straubel was there in the early days of Tesla where Elon was sleeping on the factory floor.

[00:05:08] JB himself was not getting home very often.

[00:05:11] And one of the things that they were doing was creating capacity for electric cars

[00:05:16] that a lot of people didn't think was actually going to be needed.

[00:05:18] That's a huge gamble.

[00:05:20] And ultimately, they were proven right.

[00:05:22] And having that factory actually allowed them to grow that way.

[00:05:26] And what JB is now doing is very similar to where he was years ago

[00:05:31] when they were thinking about what the world could look like for electric cars.

[00:05:35] And we talked to him about how do you scale a business for this huge future that you imagine

[00:05:40] at a time when it's a little uncertain where the electric car industry is at the moment.

[00:05:45] There are other people who feel like maybe this thing is overblown

[00:05:48] and sales are not growing at the rate that they once were.

[00:05:52] And JB is incredibly optimistic as a person.

[00:05:57] He says he's generating hundreds of millions of dollars this year,

[00:06:02] which is surprising, will be surprising to people,

[00:06:04] because the idea that there's this gold and this trash, not everybody saw that.

[00:06:09] And maybe he'll, you know, in a few years, he'll be the leading face of the big battery industry.

[00:06:15] Big battery.

[00:06:16] Tim Higgins and Christopher Mims are the hosts of Bold Names.

[00:06:20] Thanks so much for joining us.

[00:06:21] Thank you.

[00:06:22] Thanks for having us.

[00:06:23] Those were WSJ columnists Christopher Mims and Tim Higgins speaking with Danny Lewis.

[00:06:29] You can listen to the latest episode of our series, Bold Names,

[00:06:32] right here in the Tech News Briefing feed.

[00:06:35] The next episode drops on Saturday.

[00:06:37] Coming up, the CEO of collaboration software company Figma says software is boring the world.

[00:06:45] How could AI help?

[00:06:46] That's after the break.

[00:06:54] AI may be the most important new computer technology ever, but AI needs a lot of processing speed,

[00:07:00] and that gets expensive fast.

[00:07:02] Upgrade to the next generation of the cloud, Oracle Cloud Infrastructure, or OCI.

[00:07:06] OCI is the single platform for your infrastructure, database, application development, and AI needs.

[00:07:13] Do more and spend less like Uber, 8x8, and Databricks Mosaic.

[00:07:17] Take a free test drive of OCI at oracle.com slash wallstreet.

[00:07:22] Oracle.com slash wallstreet.

[00:07:28] Figma is a collaboration software company.

[00:07:32] It helps designers and product developers work together to build software products,

[00:07:37] and it's looking to incorporate AI into basically every step of the development process.

[00:07:42] That's according to its co-founder and CEO, Dylan Field.

[00:07:46] At WSJ Tech Live in October, Field joined WSJ Global Tech Editor Jason Dean

[00:07:51] for a chat about collaboration in the AI era.

[00:07:55] Here are highlights from that conversation.

[00:07:57] Are you finding any things that, as you sort of experiment and learn,

[00:08:02] generative AI is not good at in this?

[00:08:04] Like, not a good use for what you and your customers are doing?

[00:08:08] I think that what you have to start with when you're developing a product

[00:08:11] where AI is not just like sprinkled on top for marketing,

[00:08:14] but rather deeply useful for somebody to use,

[00:08:17] is you really have to start off by trying to find those areas where it can excel.

[00:08:24] And once you find an area where AI can actually be legitimately helpful to a customer,

[00:08:29] then you can worry about, okay, how much of the time does it excel?

[00:08:33] Is this like a 60% time that it works, or is it like a 99% of the time?

[00:08:37] And how do you get it from, you know, working some of the time to working all the time?

[00:08:41] You also then can worry about cost.

[00:08:43] Maybe you start off with the cost being really high.

[00:08:45] And you go, there's no way we could ever ship this to customers.

[00:08:48] It just would be too expensive.

[00:08:49] But then you can optimize that over time and make it lower too,

[00:08:52] so that your margins aren't too impacted.

[00:08:54] Or you can figure out the pricing strategy for it.

[00:08:56] Yeah, I was going to ask about the pricing strategy,

[00:08:57] because right now I think you're paying for it, basically.

[00:09:00] We're in a beta period, yeah, where we're saying, okay, this is all so early.

[00:09:03] So we'll give it away for free for a period of time.

[00:09:06] But at some point, yeah, we'll need to monetize it

[00:09:09] and make sure that we're not footing the bill for everything.

[00:09:11] Are you seeing the costs come down?

[00:09:14] Because one of the things that's interesting about Genitive AI,

[00:09:16] unlike a lot of previous, you know, the software boom of the last two decades,

[00:09:21] the scaling laws are not the same in terms of cost.

[00:09:23] Because inferencing, you know, every time you use it,

[00:09:25] you have to pay in a way that isn't true.

[00:09:28] But are you seeing that cost come down in a way that makes you optimistic

[00:09:31] that people will be able to pay for it?

[00:09:34] Gosh, this is a big question.

[00:09:35] So training and inference are both important here, right?

[00:09:38] So training is when you're setting up a model in the first place.

[00:09:41] Inference is when you're actually using the model.

[00:09:44] And on the training side, our costs are only going up

[00:09:47] as we're trying to make our product better for customers.

[00:09:50] On the inference side, I'd say you have competing things happening.

[00:09:54] So the sort of like unit cost almost is going down of inference time compute.

[00:09:58] But then the demand from our customers is going up.

[00:10:02] So I would say our costs overall are growing a lot right now.

[00:10:05] But in a way that, you know, we're pretty optimistic long term,

[00:10:09] there'll be some fixed cost part of it.

[00:10:11] Won't be all variable.

[00:10:12] Your clients are creative people, whether they be designers or developers.

[00:10:17] There's concern that Genitive AI will supplant humans in the creative process.

[00:10:23] What do you think of that?

[00:10:24] I mean, are you saying when that that's not a concern,

[00:10:27] that it's going to be augmented and not supplanting?

[00:10:31] Yeah, we actually named our feature intentionally around this.

[00:10:35] So we started off thinking of this question,

[00:10:37] and we're developing features that could generate designs.

[00:10:40] And so we launched something called First Draft,

[00:10:42] because we really think that, first of all,

[00:10:44] that's the intention is for it to be something that you start off with.

[00:10:48] And it helps you navigate the idea of design

[00:10:51] and see the different possibilities that could arise.

[00:10:54] But also, like, we want to set expectations in the right way.

[00:10:58] And it's not good enough to be called anything else

[00:11:00] other than First Draft right now.

[00:11:02] And, like, obviously, we hope we get better over time.

[00:11:05] But I currently do not see a world

[00:11:09] where there's anywhere close to people even questioning this in design.

[00:11:14] Now, in other fields, maybe.

[00:11:15] Right.

[00:11:15] But, like, in our neck of the woods, it's...

[00:11:17] Where somebody can just be like,

[00:11:18] okay, design me a new app

[00:11:21] and have these qualities

[00:11:23] and then just let the AI do it.

[00:11:25] Well, I think this gets to, like,

[00:11:26] maybe an important bigger question,

[00:11:28] which is, like,

[00:11:28] what will AI do for software in the first place?

[00:11:31] And my point of view would be that

[00:11:33] you're going to see exponentially more software,

[00:11:37] but a lot of it is not going to be very good.

[00:11:41] Why is that?

[00:11:42] Just because if everyone's, like,

[00:11:44] creating software and it's based maybe on a prompt

[00:11:47] or they're not doing much to actually differentiate it,

[00:11:50] I mean, design is the differentiator for software.

[00:11:52] And so I think the software that will stand out

[00:11:54] is a software that people are really investing time

[00:11:56] to do design work in.

[00:11:58] I think it was, like, 15 years ago

[00:12:00] that y'all published Mark Andreessen's op-ed,

[00:12:03] Software is Eating the World.

[00:12:05] Right?

[00:12:05] I basically think that's come true.

[00:12:07] Right?

[00:12:07] We've seen the exponential rise of software.

[00:12:09] No one thinks it's over.

[00:12:10] We all think it's continuing.

[00:12:10] I mean, so maybe it continues to eat the world.

[00:12:12] But if we're going to maybe, like,

[00:12:14] make a new posit today,

[00:12:16] it might be something like,

[00:12:18] software is boring the world.

[00:12:21] You know, I mean, like, we have this, like...

[00:12:23] Doesn't quite have the ring to it.

[00:12:24] No, it's not as catchy,

[00:12:25] but, like, I really do think that

[00:12:28] there's a lot of standardization

[00:12:29] when it comes to software these days.

[00:12:31] And I think the future to stand out

[00:12:33] in an ever-increasing sea of software

[00:12:35] is that people need to make their software

[00:12:38] more dynamic, more interesting,

[00:12:40] higher craft.

[00:12:41] And that's why we're really focusing on

[00:12:42] how do we empower people

[00:12:43] to do that with their design.

[00:12:46] Dylan, thank you so much for being here.

[00:12:48] Thank you.

[00:12:48] Fascinating.

[00:12:50] That was WSJ Global Tech Editor Jason Dean

[00:12:53] speaking with Figma co-founder and CEO

[00:12:56] Dylan Field at WSJ Tech Live.

[00:12:59] And that's it for a tech news briefing.

[00:13:01] Today's show was produced by me, Julie Chang,

[00:13:03] with supervising producer, Catherine Milsop.

[00:13:06] We'll be back this afternoon

[00:13:07] with TNB Tech Minute.

[00:13:09] Thanks for listening.

[00:13:10] Thanks for listening.