13 most EXPENSIVE Tech Fails of all time

[00:00] I'm a French jewel thief, but I got a

[00:01] bit of a problem. I don't have any

[00:03] jewels.

[00:05] So, I'm off for a little day trip to the

[00:07] Louvre to get some.

[00:09] The thing is though, the Louvre is

[00:11] world's most visited museum. It's home

[00:14] to humanity's rarest treasures like the

[00:16] Mona Lisa and even the French Crown

[00:18] Jewels. And so, to actually breach this

[00:20] security would require an absolute black

[00:22] ops cover of night Ocean's 11 level

[00:25] operation.

[00:28] Oh.

[00:29] >> [music]

[00:30] >> Maybe not.

[00:32] Because in October 2025, four thieves

[00:35] literally committed daylight robbery at

[00:37] the museum. They rocked up to a window

[00:39] in a furniture lift wearing high-vis

[00:41] vests

[00:43] during opening hours and they just took

[00:45] the Crown Jewels. They hopped back into

[00:47] their lift, descended comically slowly

[00:49] back down to the ground before

[00:50] disappearing away on their scooters

[00:53] at 9:30 a.m. Ooh.

[00:58] Just in time for a breakfast [music]

[00:59] croissant. See, the Louvre had a bit of

[01:02] a tech problem. Their entire

[01:04] cybersecurity system was based on

[01:05] software almost as old as me with a 2014

[01:09] audit finding that they were still using

[01:11] Windows 2000 which was well past the

[01:14] point of still getting security updates.

[01:16] And yeah, they did have CCTV. It was

[01:18] just

[01:19] facing the other direction. So, they had

[01:21] no view of this balcony being used to

[01:23] access the window. Okay, but still these

[01:25] thieves would have had to be something

[01:27] special, right? To be able to identify

[01:29] this critical blind spot. Oh, that's

[01:31] right. The password to the entire video

[01:34] surveillance network for the Louvre

[01:37] was Louvre.

[01:40] It was not a joke earlier by the way.

[01:41] This actually happened allowing them to

[01:44] drive off into the

[01:46] rush hour traffic with a hundred and two

[01:49] million dollars of jewels. The Louvre's

[01:51] president, Laurence des Cars, was so

[01:53] appalled that she tried to resign the

[01:55] very next day. And I would still [music]

[01:57] only call this a five out of ten tech

[01:59] fail. Because it got me thinking, if an

[02:01] easy to guess password and four dudes in

[02:03] a furniture lift are enough to cause a

[02:05] hundred and two million dollars worth of

[02:07] damage, then how expensive can it get

[02:09] when the mistakes get bigger?

[02:12] That's what I found out and I kind of

[02:14] wish I hadn't.

[02:15] Like,

[02:16] >> [music]

[02:16] >> you know Dyson, the company who makes

[02:18] your vacuum, your hand dryer, your hair

[02:20] dryer. Essentially, if it moves air,

[02:22] then Dyson will sell it to you for like

[02:24] $400.

[02:25] Well, in 2016, James Dyson decided that

[02:28] what he really wanted to move was

[02:30] people. So, he started secretly building

[02:32] an electric car. And look, it wasn't a

[02:35] crazy idea. Dyson specializes in

[02:38] high-performance batteries and electric

[02:39] [music] motors and that's exactly what

[02:42] an EV needs, which is why they went big

[02:44] developing a battery back that could go

[02:46] 600 miles on a single charge. We're

[02:48] still not seeing 600 miles now [music]

[02:50] and this was 2019. Not to mention a

[02:52] floating hologram heads-up display,

[02:55] seven seats [music]

[02:56] and a design that I think I'm physically

[02:58] attracted to. Oh yeah, and the seats

[03:00] were ergonomically redesigned from

[03:02] scratch because Mr. Dyson personally

[03:04] hates the lack of lumbar support in

[03:06] normal cars.

[03:09] I can kind of see his point to be fair.

[03:11] Not our most comfortable getaway.

[03:14] The slight problem is Dyson ended up

[03:16] reworking so much that they eventually

[03:19] realized just to break even they would

[03:21] have to sell each of these cars at

[03:23] today's equivalent of $275,000. [music]

[03:28] I don't think we have enough jewels. And

[03:30] it took our good friend James burning

[03:32] through 900

[03:34] >> [music]

[03:34] >> million dollars

[03:36] of his own personal money to come to

[03:38] this realization. At which point he had

[03:40] no choice but to scrap the entire

[03:42] project, go home and

[03:44] sit on a [music] chair that

[03:46] probably gave him back pain.

[03:48] Six out of ten.

[03:49] Lads, do you want some lunch?

[03:52] But sometimes the biggest cost of a

[03:54] mistake is to your reputation like it

[03:56] was with [music] Taco Bell. Because in

[03:58] 2025, this company implemented an

[04:01] AI-powered drive-thru system in over 500

[04:04] of their restaurants hoping to improve

[04:06] customer experience and um

[04:09] let's just say it wasn't very good at

[04:11] taking orders. [music]

[04:12] And what would you like to drink with

[04:13] that?

[04:16] I want a large Mountain Dew.

[04:20] And your drink?

[04:22] >> [laughter]

[04:24] >> Oh, and my absolute favorite is when

[04:26] this guy asks, "Can I get 18,000 water

[04:29] cups please?"

[04:33] >> [music]

[04:34] >> Okay.

[04:36] What can I get for you? The way that the

[04:38] system just instantly dies and hands it

[04:40] over to a human worker is class. And

[04:42] just a repost of this AI breakdown on

[04:44] YouTube is currently sitting at nearly

[04:46] 30 million views. And so, responding to

[04:49] the fact that their AI's greatest

[04:50] achievement was becoming a meme, Taco

[04:53] Bell's chief digital officer, Dane

[04:54] Matthews, told the Wall Street Journal

[04:56] that they were going to think carefully

[04:58] [music] about where not to use AI going

[05:01] forward. Gee, thanks Taco Bell.

[05:04] How did that not occur to you before?

[05:06] Having just watched McDonald's cancel

[05:08] their own AI drive-thru after it went

[05:09] viral for putting bacon on ice cream and

[05:12] nine sweet teas in one order. They keep

[05:14] trying.

[05:15] They keep failing.

[05:16] And in the process, they just keep

[05:18] funding their own funerals. Where's the

[05:20] rest? Is that all they had left? What

[05:21] about yours? I had a croissant.

[05:25] Right.

[05:26] >> [music]

[05:27] >> I should probably go then.

[05:28] Ah.

[05:30] But it's more embarrassing for Will

[05:32] Smith who used AI to replace people but

[05:35] just

[05:36] didn't even realize. Last year he posted

[05:38] a video from his music tour and

[05:39] >> [music]

[05:40] >> do you notice anything a little off

[05:41] about it? Specifically the crowd that

[05:43] has clearly been edited with AI. It kind

[05:46] of looks like Will wasn't quite

[05:47] satisfied with the turnout so decided to

[05:50] embellish it a little with some

[05:51] AI-generated concert goers hoping no one

[05:54] would notice. Big fan of the guy so

[05:55] moved by Will's music that he's wiping

[05:57] his tears through his glasses. The sign

[06:00] saying from West Philly to West Swiggy

[06:03] and especially the one claiming that

[06:05] Will helped them survive cancer. The

[06:07] idea of all of this being AI-generated

[06:10] was horrific news for Will Smith's

[06:13] let's be honest, already waning

[06:14] reputation. With every comment some sort

[06:17] of joke at his expense like unreal

[06:19] concert man.

[06:21] Literally. And Will Smith has not only

[06:23] just melted his fans' hearts with his

[06:24] concert but also melted their entire

[06:27] bodies. But the worst part, the entire

[06:29] fiasco was an accident. Turns out the

[06:31] crowds were real. You can see them in

[06:33] phone footage from the gigs and

[06:35] bafflingly in photos posted by Smith

[06:38] himself where you can directly

[06:39] cross-reference things like the Swiggy

[06:41] sign was there in his Switzerland show

[06:44] although it actually says Swizzy. So,

[06:46] most likely Will's social media team

[06:48] just used an AI tool to put together the

[06:51] highlight reel. They fed in genuine

[06:53] photos and videos but because the tool

[06:55] itself was AI, it introduced all of

[06:57] these unintended side effects. Either

[06:59] way, damage is done. Everyone thinks

[07:01] Will Smith faked a crowd to stoke his

[07:03] ego. He's the AI crowd guy now. Four out

[07:06] of ten.

[07:08] Oh, there's that.

[07:10] There's that. But speaking of men with

[07:12] image problems, you've probably seen

[07:14] Elon Musk's Tesla Optimus humanoid robot

[07:18] at this [music] point even if it was

[07:19] only as a guy in a costume. Musk has

[07:21] stated that the Optimus robot will

[07:23] eventually account for 80% of Tesla's

[07:26] value which

[07:28] got to say sounds [music] like a

[07:29] ludicrous prediction. He's also claimed

[07:31] that Tesla currently has two of these

[07:33] Optimus robots actually working. So,

[07:35] when you see one out here serving drinks

[07:37] at a product demo, you're kind of

[07:39] expecting the real deal, right? Well,

[07:44] um

[07:49] Oh.

[07:51] The fall was embarrassing enough, but

[07:52] it's actually the hand movement that got

[07:54] people talking. Look closer. Doesn't it

[07:56] look an awful lot like it's removing a

[07:59] headset kind of like the thing was

[08:01] actually being remote controlled by say

[08:04] a human operator. And Tesla have a track

[08:06] record of doing exactly this, putting

[08:09] human-operated robots out there and

[08:11] deliberately being very lax about

[08:13] letting people know that these are not

[08:15] fully autonomous. Add in the fact that

[08:17] Musk is currently chasing a one trillion

[08:19] dollar payout from Tesla where one of

[08:22] the goals that he needs to hit is to

[08:24] ship a million bots and it starts to

[08:26] make sense why he's doing way too much

[08:28] to convince you that the future is now.

[08:30] Maybe just hang on until you have a

[08:32] working prototype before showing it off.

[08:34] At least that's what Nothing did when

[08:36] they wanted to impress customers with

[08:38] the camera on their new flagship phone

[08:41] three holding a bunch of demo events in

[08:43] stores. Five photographs were shown off

[08:45] on these in-store demo units with the

[08:47] text, "Here's what our community has

[08:49] captured with the phone three." Do you

[08:50] want to know what's crazy? Zoom into the

[08:52] reflection of this one and you can

[08:54] actually see the DSLR camera, not phone

[08:57] three, that took it. And then with a

[09:00] little internet sleuthing, fans came to

[09:02] realize that in fact every single one of

[09:05] these shots was in fact a publicly

[09:07] available stock photo. Nothing

[09:08] eventually came out to claim that they

[09:10] were just placeholder images that they

[09:12] had intended to replace but that the

[09:14] units had gotten to stores before that

[09:16] happened. Malicious or not though, this

[09:17] is a pretty costly extremely avoidable

[09:20] PR moment to be having while being the

[09:22] underdog just as you release your first

[09:25] ever flagship smartphone.

[09:27] But Nvidia has definitely lost a lot

[09:30] more rep.

[09:31] And didn't have a huge amount to lose in

[09:32] the first place ever since they pivoted

[09:34] their focus towards providing graphics

[09:36] cards to giant AI companies kind of

[09:38] ignoring their original customers, the

[09:40] gamers.

[09:41] But then, instead of deciding that it

[09:43] was time to listen to the players,

[09:46] Nvidia just decided to go full on

[09:48] friendly fire with DLSS 5. So, DLSS,

[09:52] which stands for deep [music] learning

[09:53] super sampling, has been one of Nvidia's

[09:55] superpowers for a long time. It's

[09:57] basically a smart graphics technology,

[09:59] which means that instead of each

[10:01] generation having to keep doubling the

[10:03] amount of hardware they're giving you,

[10:05] they can instead upgrade the resolution

[10:07] and the frame rates in your games using

[10:09] clever machine learning tricks. However,

[10:12] the latest version, DLSS 5, goes one

[10:15] step further to upgrade even the

[10:17] lighting and the textures, too, which

[10:19] sounds fantastic, right? Who doesn't

[10:21] want to prettier game? But let's

[10:23] actually think about it for a second.

[10:24] This scene of Grace from Resident Evil

[10:26] Requiem is a scene of her heading to

[10:28] investigate the house where her mom was

[10:30] murdered in front of her eyes. But

[10:32] Nvidia's AI slop filter doesn't know

[10:35] that. It just sees a face that looks a

[10:37] little dark and traumatized, uses its

[10:40] training data, which tells it that faces

[10:42] look better when they're yassified with

[10:44] blush and eyeliner like your resident

[10:46] e-girl, but in the process completely

[10:48] flies in the face of the possibly months

[10:51] that the developers spent handcrafting

[10:53] the very deliberate way that they wanted

[10:55] her to look in the scene. Or why does

[10:57] the professor from Hogwarts Legacy need

[10:59] more wrinkles? Guys, it really feels

[11:01] like this is just AI seeing an older

[11:03] person, then going through the thousands

[11:06] of stock photos it's been trained on to

[11:08] make her more

[11:10] old. It's funny because CEO Jensen Huang

[11:12] responded to the overwhelming backlash

[11:15] arguing that it's not about just putting

[11:17] an AI filter over everything, and that

[11:18] instead DLSS 5 is meant to be integrated

[11:21] with the artist. And so it's it's about

[11:24] giving the artist the tool of AI. Which

[11:27] made it extra awkward when it came out

[11:29] that the developers of these games found

[11:31] out at the same time as the public that

[11:34] their games were being altered in this

[11:35] way.

[11:37] Five out of 10. [music] Nvidia's gamer

[11:39] cred was already in the bin.

[11:41] Now it's in hell.

[11:43] And while we're down there, there's a

[11:44] few fails so far where there's been some

[11:47] mystery as to whether or not it's been a

[11:49] deliberately perpetrated crime or not.

[11:51] Not so much the case with ransomware, a

[11:54] type of malware that locks you out of

[11:56] accessing your data so that criminals

[11:58] behind the attack can sell it back to

[12:00] you. But at least there are companies

[12:01] out there like Digital Mint who

[12:03] specialize in negotiating those ransom

[12:05] payments down.

[12:07] Right? You know, people who truly

[12:09] understand the criminal mind. [music]

[12:12] Yeah, about that. In 2023, it was

[12:14] actually employees from inside these

[12:16] companies who used this specialized

[12:19] knowledge to carry out their own

[12:21] attacks. They targeted at least five

[12:23] American firms. They stole the data and

[12:25] demanded millions in return. But this is

[12:27] where it gets truly unhinged. When those

[12:29] victims panicked and they called Digital

[12:31] Mint up for help, who did Digital Mint

[12:33] assign to the cases? Well,

[12:36] none other than one of the guys that

[12:38] attacked them, Angelo Martino, who in

[12:41] basically the human embodiment of the

[12:43] evil Kermit meme, had managed to put

[12:45] himself in a position where he was

[12:46] playing both sides and negotiating with

[12:49] himself. Must have been some tough

[12:51] negotiations because he managed to get

[12:54] all five of these companies to pay up.

[12:56] One of them even paid $26.8 [music]

[12:58] million and it was a nonprofit. What?

[13:03] Definitely a nonprofit now. Can you

[13:05] believe that just months before Digital

[13:07] Mint featured one of these guys in their

[13:09] employee spotlight? And I quote, "We

[13:12] bridge the gap between good, hardworking

[13:14] people and bad actors." They had no idea

[13:17] how right they were. And basically

[13:19] everyone lost here. The ransomed firms

[13:21] ended up paying out $75 million.

[13:24] No one's ever going to call up Mint

[13:26] again without thinking twice. And the

[13:28] three employees involved have been

[13:29] charged with up to 20 years in prison

[13:32] each.

[13:33] Six out of 10 fail.

[13:35] But where it goes beyond funny into just

[13:37] straight up terrifying is when a man

[13:39] named Sami Asdufal developed an app to

[13:42] allow him to control his DJI Romo robot

[13:44] vacuum using a PS5 controller simply cuz

[13:47] he said sounded fun. That's not the

[13:49] scary part. Unfortunately, Asdufal

[13:52] didn't realize his own coding [music]

[13:54] power. The custom remote control app he

[13:56] built pretty quickly using Claude AI

[13:58] code accidentally granted him control of

[14:01] over 7,000 DJI robot vacuum cleaners

[14:04] across 24 countries the second it

[14:07] connected to DJI servers. How? Because

[14:09] instead of DJI encrypting the data from

[14:12] its devices like they should have, it

[14:13] was literally laid out to him in plain

[14:16] text. Zero authentication required,

[14:18] allowing him complete access. So, okay,

[14:21] he could now vacuum anyone's house on a

[14:23] whim. That feels like a Doofenshmirtz

[14:26] level scheme. Evil, but ultimately

[14:28] harmless. Until you realize that these

[14:30] robots all had cameras and microphones,

[14:33] and that he also happened to have access

[14:35] [music] to the live feeds from them. Not

[14:37] to mention the floor plan of each house

[14:39] that the vacuums had mapped out with all

[14:40] of their sensors and their location,

[14:43] too, via their IP addresses. Thankfully,

[14:45] he was a nice guy about it. He

[14:46] immediately reported the flaw to DJI and

[14:49] they immediately rewarded him with a 30K

[14:51] payout. But just imagine how

[14:53] catastrophic this could have been if he

[14:56] wasn't a nice guy. Remember, Asdufal

[14:58] didn't hack into DJI servers or do

[15:00] anything complex. He simply became god

[15:03] of all robo vacs by accident. And it's

[15:05] just mad to think that someone can so

[15:08] easily stumble into so much personal

[15:10] data from you. At least when you're

[15:12] using Surfshark VPN, our sponsor, anyone

[15:14] who might be snooping isn't [music]

[15:16] getting anything useful cuz your

[15:18] identity is masked. just made it even

[15:20] crazier by launching this brand new

[15:22] custom standard called Dosos. Instead of

[15:24] cramming everyone's data into one shared

[15:26] tunnel like a traditional VPN, Dosos

[15:29] gives you your own dedicated private

[15:31] data lane. And the reason that I bang on

[15:33] about Surfshark is because it's like the

[15:35] one deal in tech that feels like a

[15:39] steal. With the code boss, you can

[15:40] literally get Surfshark for $1.68 a

[15:43] month cuz they're celebrating their

[15:45] birthday. And that would cover you plus

[15:47] an entire crowd at a Will Smith concert

[15:50] if you wanted. Now, at least Sami had

[15:52] the decency to feel bad about his

[15:54] accidental data theft. Can't say the

[15:56] same for ChatGPT, which is meant to

[15:58] revolutionize search by accessing

[16:00] everything on the internet and bringing

[16:02] it to you. Now, that would require

[16:04] having the rights to a lot of content

[16:06] that would end up expensive and

[16:08] time-consuming to acquire. So,

[16:09] naturally, OpenAI just ignores that

[16:11] part. Why?

[16:13] Go fast and break things, of course. Who

[16:15] needs the law when you have a

[16:16] ridiculously ballooned valuation? So,

[16:18] it's a shock to absolutely no one then

[16:21] that a group of authors and publishers

[16:23] are suing OpenAI for copyright

[16:24] infringement. But it's what's happened

[16:26] as part of their investigation that's

[16:28] been the dramatic upset because the

[16:30] group managed to acquire leaked Slack

[16:32] messages and emails from OpenAI in which

[16:35] their employees openly discussed the

[16:37] mass deletion of two data sets the AI

[16:39] was trained on that they knew consisted

[16:42] of pirated books. Hilariously named

[16:45] books one and books two, by the way, in

[16:47] case it was unclear. So, not only did

[16:49] they have other people's pirated data,

[16:52] but they knew full well that what they

[16:54] were doing was wrong and tried to

[16:56] dispose of the evidence. So, a New York

[16:59] District Court has now ordered OpenAI to

[17:01] hand over those messages. And if those

[17:03] messages demonstrate willful

[17:05] infringement, and I don't really see how

[17:07] they couldn't, this could take the

[17:09] damages anywhere from $750 per piece of

[17:13] stolen work to possibly $150,000

[17:16] per work. An insane amount on the scale

[17:19] of data that these guys are working

[17:21] with. We don't know the total fines they

[17:23] could face yet, but we can get some idea

[17:25] from a recent lawsuit against Anthropic

[17:27] for similar copyright infringement,

[17:29] which saw them settling for $1.5

[17:32] billion. Dollars.

[17:34] Settling.

[17:35] Seven out of 10. But while the exact

[17:37] consequences of that are still up in the

[17:39] air, OpenAI has recently had an even

[17:41] bigger oopsie that's already cost them

[17:44] some very real, very large dollars. See,

[17:47] in 2025, the company announced a noble

[17:49] new mission that doom scrolling wasn't

[17:52] bad enough as it was, and that what

[17:53] humanity really needed was an entirely

[17:55] new short-form video app called Sora

[17:58] that lets you doom scroll content that

[17:59] was entirely [music]

[18:00] AI generated. A

[18:03] slop talk, if you will.

[18:04] >> [music]

[18:04] >> And to kick off the fun while

[18:06] downplaying the dystopian impending

[18:08] threat of anyone being able to create a

[18:10] deepfake in like two clicks, Sam Altman

[18:13] gave all users global permission to

[18:15] create videos using his own likeness,

[18:18] which of course immediately backfired

[18:20] with a litany of embarrassing videos

[18:22] mocking the guy. Like this one of him

[18:24] begging for GPUs at a doorbell camera. I

[18:26] can't train anything. Please, if you

[18:27] have anything, A100s, 3090s, I'll take

[18:29] them.

[18:29] >> Physically stealing art from Studio

[18:31] Ghibli. Give them back. NOPE, TOO LATE.

[18:33] HEY, COME BACK HERE. FREE YARD, BABY. OR

[18:36] THIS ONE of Altman hosting a Hunger

[18:38] Games style competition forcing

[18:40] contestants to fight over literal slop.

[18:43] And then of course this, which I present

[18:45] without comment.

[18:47] Meow.

[18:48] Turns out Super Sam did this to himself

[18:51] for absolutely nothing because Sora shut

[18:54] down just six months after starting. Let

[18:56] me try and put into perspective just how

[18:58] much of a flop this slop was. Sora

[19:01] generated in total $2.1 million in

[19:05] revenue.

[19:06] Sora was costing them at peak usage

[19:09] around $15 million per day

[19:12] to run all while their user base was

[19:15] collapsing under their feet with a 66%

[19:18] drop in just the first [clears throat]

[19:20] 90 days. And as if that wasn't already

[19:22] enough damage, OpenAI also, in the

[19:24] process of losing Sora, lost an

[19:26] investment deal with Disney. Disney was

[19:28] going to pay OpenAI for a stake in the

[19:30] company, which would have also given

[19:32] Sora users access to 200 plus characters

[19:35] from across the entire Disney universe.

[19:37] How much was Disney going to pay them?

[19:40] $1 billion.

[19:44] Oops.

[19:45] Now, you already know that in 2022, Elon

[19:47] Musk purchased Twitter for mountains of

[19:50] money.

[19:51] We've been living with the consequences

[19:53] ever since. But there is one consequence

[19:55] that you might have missed. That Musk is

[19:57] now being sued by the investors who

[19:59] originally put their money behind that

[20:01] deal. And the reason is one of the

[20:03] funniest things I've ever seen in a

[20:05] courtroom. So, do you remember when Musk

[20:06] was trying to wiggle out of buying

[20:08] Twitter? He'd already committed at this

[20:09] point, but he was trying to pull away

[20:11] because he said too many of its users

[20:13] were actually bots, not people. And by

[20:16] the way, he said this on Twitter,

[20:18] publicly.

[20:19] In tweets. Well, those tweets tanked

[20:22] Twitter's stock price. That made all of

[20:24] these investors panic. They believed

[20:26] him. They assumed that the whole thing

[20:27] was falling apart. And so, they

[20:29] hurriedly sold the shares they had,

[20:31] losing millions compared to what they

[20:33] paid for them. But the deal wasn't off.

[20:35] Twitter took Elon to court and forced

[20:38] him to buy Twitter anyway at the

[20:39] original agreed price of $54.20 per

[20:43] share. But all those investors who sold

[20:46] their shares at a loss,

[20:47] they never got their money back, which

[20:49] leads us to now, where a California jury

[20:51] has declared Musk liable for misleading

[20:53] them with his own tweets. Tweets which

[20:56] he himself described under oath as

[21:00] stupid tweets. You just can't make this

[21:02] stuff up. The man who bought Twitter to

[21:04] protect free speech is being sued for

[21:07] what he freely said,

[21:09] stupidly, for up to $2.6 billion. And if

[21:13] you think that's a lot of money, then

[21:15] you're not ready for the metaverse. Do

[21:17] you remember when Facebook rebranded to

[21:19] Meta out of nowhere and made a massive

[21:21] bet on our near future becoming almost

[21:23] exclusively virtual reality? The hub for

[21:25] which was meant to be Horizon Worlds, an

[21:27] online VR world where you can hang out

[21:29] with all your other friends who

[21:30] definitely have a Quest headset.

[21:33] Now, because they were trying to take VR

[21:35] from merely something that people used

[21:37] to play games into this alternate

[21:39] reality where we'll all one day live and

[21:41] work as well as play, it was important

[21:44] to have the infrastructure for users to

[21:46] visit different worlds and create their

[21:48] own for

[21:50] whatever use they might want. Like for

[21:51] instance, taking a fake selfie in front

[21:53] of a fake 240p Eiffel Tower while trying

[21:57] not to let the fake existential dread

[21:59] creep in too much.

[22:01] Wait, no. Dread was real.

[22:04] Well, it seems like Meta are starting to

[22:05] catch up with the rest of us who already

[22:07] know that no one's interested in hanging

[22:09] out in VR, whether it gives us legs or

[22:11] not. Because this year, they announced

[22:13] that they were removing Horizon Worlds

[22:14] from the Quest app store, deciding

[22:16] instead to focus 100% of their effort on

[22:19] mobile instead. Cuz yes, this is exactly

[22:22] what the phone user base has been

[22:24] yearning for. But then, a glimmer of

[22:26] hope for the roughly four remaining

[22:29] Horizon Worlds fans, Meta did a U-turn

[22:31] and announced that in fact, no, we're

[22:33] going to be keeping VR access to the

[22:34] platform after all. Just, according to

[22:36] them, to support the fans who've reached

[22:39] out.

[22:40] Must have been a long day going through

[22:41] all of those emails. But Horizon Worlds,

[22:43] as embarrassing as it is, is just one

[22:45] tiny part of the metaverse's failure.

[22:47] Because

[22:48] amidst the fallout, we've also come to

[22:50] learn the total amount of money that the

[22:52] whole metaverse project has lost Meta.

[22:55] And

[22:56] you're going to want to sit down for

[22:58] this one. We've seen millions this

[22:59] video. We've even seen a couple of

[23:00] billion. But Meta burned through $80

[23:04] billion. That's pretty much what it

[23:07] would cost to end world hunger for a

[23:09] year.

[23:10] What did it get spent on?

[23:12] This.

[23:14] Cheers Zuck.

[23:17] I got to get out of here.