How Big Tech really feels about the EU’s crackdown

With the EU’s Digital Services Act now being enforced, 19 of the largest online platforms are subject to the most stringent tech regulations ever imposed by the Western world. These Very Large Online Platforms, or VLOPs, as the EU calls them, all have at least 45 million monthly users in the bloc. That deems them “systemically relevant,” according to EU Commissioner Thierry Breton.

EU citizens will observe the impact of the DSA immediately. The big social networks, including Facebook, Instagram, Snapchat, and TikTok, now let people in the EU opt out of having their feeds personalized. They have to be more clear about how and why they moderate a wider range of content, and they have to let people in the EU appeal those decisions. And all of the ads these platforms run in the EU will be browsable in each company’s respective ad library.

Most of the VLOPs have put out press releases touting their compliance with the DSA and the level of effort that went into achieving it. Meta president Nick Clegg’s blog post boasts that the company “assembled one of the largest cross-functional teams in our history, with over 1,000 people currently working on the DSA.” Google’s post lists the “significant investments” it has made to comply with the new rules, though it hasn’t shared a headcount number. The EU government, meanwhile, is expecting to have only about 230 staff members actually enforce the DSA, according to The Wall Street Journal.

No spokesperson for a VLOP will say this on the record, but I will: By essentially punishing companies for reaching a certain size, the DSA further entrenches the power of the very tech platforms it’s trying to bring to heel. The VLOPs know this. And they secretly like it.

Don’t believe me? Here’s how someone who has been helping lead DSA compliance for a big tech company put it when I asked how they were feeling this week:

“It was a shit load of effort to do all of it, which is a real shame since the EU could have done things that would have been actually meaningful. Instead, they just wasted Big Tech’s time on stuff that won’t actually help people and meaningfully raised the bar for complying with the law for any platform that reaches VLOP status. If I’m an emerging startup, I’m terrified.”

I should have put the news of David Limp’s “retirement” from Amazon in last week’s issue in quotes. The whispers I’ve since heard are that the devices and services chief was pushed out, despite CEO Andy Jassy’s note to employees framing it as Limp deciding to retire.

There have been plenty of signs that this was coming: The first was Jassy’s gutting of Limp’s org through layoffs in November. The second was the highly unusual move he made last month, when he took Rohit Prasad, SVP and head scientist for Alexa, out from Limp’s list of direct reports to lead generative AI across Amazon and report directly to him.

Unlike Jassy’s memo announcing Limp’s departure, the note that Limp shared with employees doesn’t mention retirement at all. Instead, he admits that he doesn’t know what he’s doing next, only that he knows it “won’t be in the consumer electronics space.” That doesn’t sound like retirement to me!

Nearly eight years after its initial release, Alexa never became the “fourth pillar” of Amazon’s business that Jeff Bezos hoped it would be. Instead, Limp’s division has been a money pit, which is no longer okay as Jassy seeks to cut unprofitable businesses across the company. Ken Washington, Amazon’s VP of consumer robotics and one of Limp’s reports, recently left, as did Ring founder Jamie Siminoff.

It’s clear that Amazon was caught flat-footed by the generative AI boom, though Limp recently told The Information that “I don’t feel like we’re behind” on AI. (The only thing I still use the Alexa in my kitchen for is a cooking timer.) The Astro smart home robot still isn’t available to the general public after initially debuting in 2021. Meanwhile, his org’s competitor to Starlink, Project Kuiper, has yet to launch a single satellite. Whoever his successor is that Jassy names in the coming weeks has their work cut out for them.

Apple has been bringing app developers into its offices to try the Vision Pro. I got a kick out of this post by Apple rounding up some reactions from friendly developers who have gotten to try it, including one who says he audibly gasped when he saw his app running in the headset for the first time.

While most developers have to physically go into one of Apple’s labs to try the Vision Pro, a select group has been able to take headsets home to spend more time building for the device. After talking with one such developer, I’ve confirmed the suspicion I had after my own Vision Pro demo back at WWDC: key aspects of the device don’t work yet, namely the front-facing display that is supposed to show the wearer’s eyes as they move.

That explains why no Apple execs have been photographed wearing the Vision Pro yet, and why even the accompanying photo for the company’s post this week on early developer reactions only shows the headset being worn from the side. Siri also doesn’t work on the devices that have been loaned to developers, I’m told. Has Apple ever let people from the outside world use, much less take home, a device that is this unfinished?

Apple is, of course, very aware of all this and doing its best to keep non-authorized people from using or dissecting the headset. The units that have been sent home with developers have AirTags physically attached to them to track their locations, and a contraption on the temple is designed to keep someone from physically opening the device up and inspecting its innards.

“A new computing era has begun.” - Nvidia CEO Jensen Huang during the company’s second quarterly earnings call.

If this year’s bonanza rush for AI chips can be summed up in one stat, it’s this: 171 percent. That was Nvidia’s revenue growth last quarter from the same time last year for its data center business, which houses its AI chip sales. As I explained in a recent issue, those chips — specifically the H100 — are the tech industry’s most sought-after resource at the moment.

I’ve been reading Wall Street analyst reports on Nvidia’s earnings beat and its monster guide for Q3, which is roughly $4 billion above estimates. That kind of forecast rarely, if ever, happens with mature, public companies. It’s indicative of how quickly the demand for generative AI products has ramped and that, at least for now, Nvidia is essentially the only game in town.

Huang didn’t share any specifics about exactly how much Nvidia plans to accelerate production of the H100 going forward, so for now, I’ll leave you with this report from the Financial Times, which says to expect shipments of 1.5 to 2 million units next year, up from roughly 500,000 this year.

A roundup of what else is going on in and around tech companies this week:

We just announced that Casey Bloys, the chairman and CEO of HBO / Max, is speaking at the Code Conference in late September. We’ll also have Cristóbal Valenzuela, the CEO of Runway, Rita Popova, chief product officer at Replika, and others for a spotlight session on generative AI.

Ping me if you’re interested in attending, or fill out the application here.

I’m off next week but will be back with my mailbag issue after Labor Day.

Thanks for subscribing.