• last year
The deal was one of several challenges facing the federal business of the $7.3 billion AI startup, whose CEO Alexandr Wang has become an outspoken advocate of American national security amid the threat of China. One expert called the optics “horrible.”

0:00 Introduction
1:05 About Alexandr Wang's Contributions To AI
4:02 Scale AI, Where Are They Now Business Wise?
7:44 What Does The Future Hold For Scale AI?
9:42 How The US Views AI?
13:54 The AI Climate In Silicon Valley

Read the full story on Forbes: https://www.forbes.com/sites/davidjeans/2024/02/02/scale-ai-tiktok-bytedance-security-concerns/?sh=155f40a52733

Subscribe to FORBES: https://www.youtube.com/user/Forbes?sub_confirmation=1

Fuel your success with Forbes. Gain unlimited access to premium journalism, including breaking news, groundbreaking in-depth reported stories, daily digests and more. Plus, members get a front-row seat at members-only events with leading thinkers and doers, access to premium video that can help you get ahead, an ad-light experience, early access to select products including NFT drops and more:

https://account.forbes.com/membership/?utm_source=youtube&utm_medium=display&utm_campaign=growth_non-sub_paid_subscribe_ytdescript

Stay Connected
Forbes newsletters: https://newsletters.editorial.forbes.com
Forbes on Facebook: http://fb.com/forbes
Forbes Video on Twitter: http://www.twitter.com/forbes
Forbes Video on Instagram: http://instagram.com/forbes
More From Forbes: http://forbes.com

Forbes covers the intersection of entrepreneurship, wealth, technology, business and lifestyle with a focus on people and success.

Category

🤖
Tech
Transcript
00:00 [MUSIC PLAYING]
00:02 Hi, everyone.
00:03 I'm David Jeans.
00:04 I'm a senior writer at Forbes on the tech team.
00:07 I cover AI, national security, and defense.
00:11 I am joined from San Francisco, Kenrik Kai,
00:15 he's a senior reporter on the tech team as well.
00:18 And we're going to talk today about a story we published just
00:21 recently about a major AI company called Scale AI.
00:26 They've raised-- they're valued more than $7 billion.
00:30 And they've been receiving a lot of hype and interest
00:35 recently about some of the work they're doing.
00:37 But together, we took a look at what the company's
00:40 been doing in the federal space and trying
00:43 to work with the US government.
00:45 So I think it would be great to hear from you, Kenrik,
00:48 on how we first became interested in scale.
00:52 And from there, we can talk more about why
00:58 they decided to target contracts and generate revenue
01:02 from the US government.
01:06 Yeah, absolutely.
01:06 Thanks, David.
01:07 So been keeping an eye on scale for a couple of years now.
01:11 And we put their CEO, Alexander Wang, on our magazine cover
01:16 last year because scale has emerged
01:20 in the last couple of years as one of the most vocal
01:22 and high-profile startups building
01:26 in the artificial intelligence space.
01:29 CEO Wang was, at one point, the youngest self-made billionaire
01:34 as we pegged him on our billionaires list.
01:37 And the company has followed a couple of different AI waves
01:41 to increasing revenue.
01:44 A couple of years ago, it was with self-driving cars,
01:48 helping label the data to allow those cars
01:52 to drive autonomously.
01:53 And recently, we're hearing that their revenue has continued
01:57 to skyrocket from about $250 million
02:01 in annual recurring revenue last year to north of $700 million.
02:05 So it's a company that continues to position themselves
02:09 as one of the key infrastructure suppliers for the AI
02:15 ecosystem.
02:16 But in the course of reporting our magazine story last year,
02:21 and I think this time around as well with you, David,
02:24 finding that there is more than meets the eye in terms
02:27 of what the company, I think, kind of projects
02:31 as their image versus what's going on on the ground.
02:34 So you might talk about some of the findings
02:38 that we found on the federal side this time around.
02:41 Yeah, I think also just to step back on that,
02:44 I think the reason we really started
02:46 looking at the federal business and their efforts
02:48 to sell to the US government and especially the military,
02:54 this was interesting to us because the company,
02:58 and especially CEO Alexander Wang,
03:01 have often talked about the need for US
03:04 to be better in the AI space and win against China.
03:10 And there's been a lot of discussion about China
03:12 and the race between the US and China on technology supremacy,
03:17 really.
03:17 And I think given this dialogue around this space,
03:24 I think that it was interesting for us
03:28 to take a closer look at what scale
03:31 is doing to advance the US national interests there.
03:36 And what we really discovered was
03:39 that there have been some efforts by the company where
03:42 they've run into some conflict between balancing
03:46 the needs of the US federal government
03:48 and also its commercial interests.
03:51 And I think, Kenric, maybe it'd be
03:53 good if you could talk a bit about our big finding.
03:56 But there's a lot more to our story
03:58 than beyond this headline.
04:02 Yeah, I think that's a really good point.
04:05 You're right on.
04:06 CEO Wang has, in the last several years,
04:10 talked about the US and China as largely a two-pronged war
04:16 that he has dubbed the AI war.
04:20 And has spoken about China's evolving artificial intelligence
04:25 capabilities as a national security concern.
04:30 He wrote on Twitter at some point last year
04:34 that there is a non-zero chance that World War III has
04:39 since already commenced.
04:42 And so, yeah, increasingly scale since a trip
04:48 that he mentioned to me during my magazine story,
04:51 as part of my magazine story interview last year,
04:54 in late 2018, took a trip to China
04:59 and visited several of the offices
05:02 of high-profile Chinese tech companies,
05:05 including TikTok's parent company, ByteDance.
05:09 And what he told me then was that he'd
05:11 come out of that experience kind of stunned
05:19 by the developments in AI that were happening in China.
05:24 And that that had spurred him to launch the federal business
05:29 arm of Scale AI to help the US government in advancing
05:37 their AI capabilities.
05:41 So in terms of that backdrop, I think
05:44 that makes what we found for our story particularly interesting,
05:48 which was that at some point in 2022, late 2022,
05:56 Scale had advanced contract talks
06:00 with TikTok, which is owned by the Chinese company ByteDance.
06:07 And that Wang had pushed this deal forward,
06:10 despite protests from several of his key executives
06:14 and people on the federal team that he had built out at Scale.
06:18 A lot of it related to the national security concerns
06:21 that we were alluding to earlier.
06:24 And this sort of--
06:27 let's say it's an interesting dynamic.
06:29 He comes away from a trip to China
06:32 with one view of why it's important to support the US
06:37 mission, and then pursues a contract with a company
06:43 that he'd visited while in China.
06:46 We spoke to a top former US government counterintelligence
06:50 expert, who we quoted in the story.
06:53 And he basically underlined the concern here,
06:56 which is that from a public perspective,
07:00 this is a really big no-no.
07:02 Because at the time, the US government
07:06 was incredibly skeptical and scrutinizing TikTok.
07:10 And it seemed to be that if Scale
07:13 wanted to maintain its federal business
07:16 and working with the US federal government,
07:18 it had to produce an image that it was responsible
07:23 and didn't have any conflicts of interest or anything
07:27 with any Chinese-backed companies.
07:29 So this was a rather interesting finding.
07:33 And after-- Scale-- it was actually
07:38 after Forbes published a story about TikTok.
07:42 And it's a parent company.
07:44 They had been planning to surveil Forbes reporters,
07:48 American citizens.
07:49 It was after that story was published in late 2022
07:53 that Scale executives and Alexander Wang
07:56 agreed that they should end conversations and negotiations
07:59 with TikTok.
08:00 So it was a series of events that
08:03 happened about 18 months ago.
08:07 However, our reporting found that we really--
08:12 that there was a lot more to the federal business
08:14 that has occurred since then.
08:15 I think, Kenric, you spent a lot of time
08:18 looking at their most recent product, Donovan.
08:21 If you could tell us a little bit about that
08:23 and how that has also raised some red flags for people
08:28 in the military sector.
08:32 Yeah, absolutely.
08:35 Yeah, I think Scale's federal business is something
08:39 that they launched with their first contract in 2020.
08:44 So it's been about four years now.
08:47 And as the company has continued to evolve,
08:52 our understanding is they're trying
08:54 to evolve this federal business as well in line with that.
08:57 And that's come with some challenges.
09:00 We reported that last year the revenue hit about $750 million
09:06 in annual recurring revenue.
09:08 But a large proportion of that comes
09:10 from the commercial side of things.
09:13 So I think, going back to TikTok,
09:15 that speaks also to the point that I
09:18 think that motivated the company's decision
09:21 to pursue that TikTok contract in the first place,
09:24 was that it was a potential large revenue stream.
09:29 But I think you mentioned Donovan.
09:31 And Donovan is their new generative AI
09:35 platform for the federal sector that they
09:38 announced middle of last year.
09:40 And our understanding, talking to several former employees
09:45 at the company and people in the industry,
09:47 is that this is kind of their big play, big bet,
09:51 to really have a standard software
09:56 offering for the federal business.
10:00 Previously, Scale was relying a lot
10:02 on individual one-off contracts with different parts of the US
10:08 government.
10:09 And oftentimes, what that entailed
10:11 was having to create bespoke products
10:17 for a specific potential customer's concerns or needs.
10:23 And so Donovan is this generative AI platform
10:26 that leverages various AI models, including
10:30 ChaiGBT, to help government officials and people
10:35 in the federal sector to answer questions that
10:39 are meant to be more attuned to federal and government style
10:48 needs.
10:50 And so Scale has been building out
10:51 the product for the last several months since its launch in May.
10:57 And what we found is that they have,
11:01 I think, to the concern of some employees
11:04 internally, but also in demoing the product to people
11:08 in the government, run into some challenges in terms
11:12 of overstating the product's capabilities
11:15 and just really bringing the product to production.
11:19 And David, if you want to talk about some of these episodes
11:22 that have happened between Scale and their attempts
11:25 to sign some of these deals with folks in government.
11:30 Yeah, there was two anecdotes that really
11:35 underscored this point.
11:38 In the initial launch of Donovan,
11:41 which they announced in a press release last May,
11:47 they had announced an agreement with a division of the Army.
11:51 And in that press release, they said
11:54 that Donovan was going to be used for critical use cases.
11:58 They said that it had been--
12:01 there was other specifics of that press release that just
12:04 simply were not accurate.
12:07 And the feedback-- what we learned
12:09 was that the head of that Army unit
12:13 had actually contacted the company
12:15 to express concern about the fact
12:17 that the agreement had been misrepresented publicly
12:21 in a way that overstated what was actually being done
12:27 as part of that agreement.
12:28 And then we also learned in a separate meeting
12:32 with another division of the military, US Central Command,
12:37 that Alex Wang had been in a demo of the Donovan product,
12:42 and that had actually not gone according to plan.
12:48 And the feedback basically was, this
12:50 is just not ready for production.
12:52 So I think what the sort of broader takeaway
12:56 here is from the fact that there's
12:58 been these issues within the federal business,
13:01 the broader takeaway is definitely
13:02 that for commercial customers and big customers,
13:09 they see a company like Scale working with the intelligence
13:12 agencies, with the military.
13:14 That is a big stamp of approval and credibility to this company.
13:18 And that's why they've pursued the federal business so
13:22 strongly.
13:24 So this is why it does matter, is
13:27 because it's such a crucial part of this company's growth.
13:31 And likewise, they also see that helping the US government
13:38 is something that is important to them.
13:39 So I think that these are definitely some challenges
13:43 that we were able to document.
13:44 And I think it adds an interesting sort of angle
13:50 to what's really going on underneath the hood at Scale.
13:54 Yeah.
13:54 Yeah.
13:55 I think that's right.
13:56 I think it really underscores just the general tension that
14:00 occurs that we've seen in a couple of different cases.
14:04 But with a Silicon Valley tech company
14:09 attempting to make a foray into Washington, DC,
14:12 and landing government businesses
14:16 is just different from landing deals with software startups
14:21 and other tech companies.
14:24 So there are different timelines for signing government
14:29 contracts.
14:30 And just the availability of contracts
14:33 is very different in the federal sector
14:35 versus in the private sector.
14:38 So I think the TikTok episode speaks to that point.
14:40 Donovan speaks to that point.
14:43 One of the former employees that we had spoken to
14:45 for the story had told us how Wang had, quote,
14:50 unquote, "disregarded our concerns"
14:53 in terms of the TikTok deal and said that, quote, unquote,
14:56 he had to focus on long-term goals
14:58 of both sides of the company.
15:00 And the TikTok deal would help that.
15:02 So there are just different tension points.
15:04 I think there are several companies that have successfully
15:09 made the foray into DC.
15:12 There's SpaceX.
15:13 There's Palantir.
15:14 There's Nandoril.
15:15 And it is a very potential lucrative revenue stream,
15:19 as you've mentioned, David.
15:20 But scale is, I'm trying to understand,
15:24 is still kind of in the early stages of getting there.
15:27 And one of the other big examples
15:30 is that they announced this big $249 million
15:36 award with the Department of Defense,
15:40 I think about two years ago now, meant
15:43 to add AI capabilities for the DoD.
15:47 But so far, none of that money has been paid out
15:51 to the company.
15:52 And it's unclear how much will ultimately
15:56 be paid out out of that $249 million.
16:00 Thanks for joining me today, Kenry.
16:02 [AUDIO OUT]
16:06 [AUDIO OUT]
16:09 [BLANK_AUDIO]

Recommended