WEBVTT
1
00:00:00.331 --> 00:00:06.115
Hi, and welcome to Technically Creative's Week in Review, brought to you by Kubrick Labs, the AI and IT partner for creative companies.
2
00:00:06.596 --> 00:00:15.882
So we're going to do it a little bit differently this time because we've heard some feedback from people that they prefer to get more of the news up front before we go into the deep dive and editorial stuff.
3
00:00:16.362 --> 00:00:20.723
Obviously, there's news within the editorial deep dive, but it's a little bit different.
4
00:00:21.123 --> 00:00:23.124
So we're going to try it this way this week.
5
00:00:23.204 --> 00:00:24.184
So I hope you like it.
6
00:00:24.344 --> 00:00:26.025
Again, always open for feedback.
7
00:00:26.045 --> 00:00:26.865
So just let us know.
8
00:00:27.005 --> 00:00:34.907
If you haven't been here before, every week I spend about 20 minutes walking you through the most important news from the increasingly tangled worlds of tech and creativity.
9
00:00:35.067 --> 00:00:38.488
And what we're trying this week is first the quick hits, and then I get into the bigger stuff.
10
00:00:39.048 --> 00:00:41.589
And this past week, honestly, you missed some weird shit.
11
00:00:42.029 --> 00:00:44.729
A chatbot company with 12 million users went bankrupt.
12
00:00:44.909 --> 00:00:48.870
And Disney, which sent it a cease and desist letter last year, is now bidding to license the tech.
13
00:00:49.450 --> 00:00:55.031
Getty Images, the people who own every photo of a businesswoman laughing at a salad, is going through Chapter 11.
14
00:00:55.651 --> 00:00:58.852
Netflix made billions of dollars, yet still cut thousands of jobs.
15
00:00:59.012 --> 00:01:01.552
Anthropic told investors it's profitable, sort of.
16
00:01:01.932 --> 00:01:04.893
And only if you look at it just the right way while squinting.
17
00:01:05.433 --> 00:01:09.698
Meta predictably rewrote terms, and publishers are pissed off about it.
18
00:01:09.798 --> 00:01:15.225
Hollywood's three biggest agencies got together and asked Congress to make Netflix buy half of its shows from somebody else.
19
00:01:15.706 --> 00:01:17.648
And Elon Musk made a car fly!
20
00:01:17.808 --> 00:01:18.829
Again, sort of.
21
00:01:19.009 --> 00:01:19.290
Did it?
22
00:01:19.530 --> 00:01:20.251
Was it a real car?
23
00:01:20.451 --> 00:01:20.772
I don't know.
24
00:01:21.832 --> 00:01:25.134
So all in all, completely normal week for the world we're living in.
25
00:01:25.314 --> 00:01:29.955
And just 10 days ago, the stock market hit an all-time high, so everything's fine, right?
26
00:01:30.616 --> 00:01:32.577
We'll get to that, but let's start with the news.
27
00:01:33.137 --> 00:01:37.579
Meta is doing their most predictable thing, changing the T's and C's and hoping nobody noticed.
28
00:01:37.979 --> 00:01:38.539
But they did.
29
00:01:38.939 --> 00:01:44.021
Meta has updated its commercial terms, taking effect on October 30th, and publishers have noticed.
30
00:01:44.361 --> 00:01:48.563
The new language says that if a business uses Meta's products, which many do for advertising—
31
00:01:48.783 --> 00:01:56.051
Meta may then use any content from that business's website to provide, develop, improve, research, and train its products.
32
00:01:56.372 --> 00:02:01.337
And if you don't like it, the terms helpfully suggest you stop using Meta's products for business.
33
00:02:01.818 --> 00:02:02.699
There's no opt-out.
34
00:02:03.159 --> 00:02:09.626
Legal experts quoted by Neiman Labs say this could give Meta a contractual basis to use publishers' content to train AI—
35
00:02:10.167 --> 00:02:12.169
And trust me, this won't stop with publishers.
36
00:02:12.449 --> 00:02:15.192
They've just picked the weakest calf in the herd to pick on.
37
00:02:15.412 --> 00:02:17.434
Of course, Meta says nothing's changed.
38
00:02:17.474 --> 00:02:20.677
The language has just been tidied into one place.
39
00:02:20.917 --> 00:02:25.961
Its crawlers won't behave any differently, and publishers can still block them with robots.txt files.
40
00:02:26.001 --> 00:02:31.586
And everybody should know about robots.txt files, but publishers are not exactly a very tech-savvy bunch.
41
00:02:32.287 --> 00:02:41.662
The fight is over what the terms allow, but it's a little bit like your landlord saying, relax, I've just consolidated all the clauses about letting myself into your apartment into one convenient clause.
42
00:02:41.882 --> 00:02:44.766
And the clause is a please knock sign on a door with no lock.
43
00:02:45.601 --> 00:02:48.302
If you're a publisher, Meta is one of the biggest sources of traffic.
44
00:02:48.322 --> 00:02:49.442
You can't really walk away.
45
00:02:49.882 --> 00:02:55.363
So agree or leave is a pretty clever way of getting consent from people who can't afford to say no.
46
00:02:55.623 --> 00:02:58.504
The copyright war so far has mostly been fought in courtrooms.
47
00:02:58.664 --> 00:03:04.986
And this front, the same as Meta has always done, is opening up in the terms of service, which is the one place on earth nobody reads.
48
00:03:05.246 --> 00:03:06.866
So expect Meta to keep doing this.
49
00:03:06.886 --> 00:03:09.907
And if you're not in publishing, expect them to do it to your industry soon.
50
00:03:10.467 --> 00:03:13.288
And then Fox's AI is now buying actual ads.
51
00:03:13.728 --> 00:03:21.011
Back in June, Fox showed off something called Ad Studio, which it billed as the TV industry's first end-to-end agentic ad platform.
52
00:03:21.211 --> 00:03:28.854
This week, it said it's now running live linear TV campaigns with WPP, Omnicom, Horizon, Stagwell, and Simulmedia.
53
00:03:29.174 --> 00:03:32.936
In English, an agency's AI talks directly to Fox's AI.
54
00:03:33.356 --> 00:03:36.477
They work out what ad inventory is available and compete.
55
00:03:36.617 --> 00:03:38.558
And that completes parts of the media buy.
56
00:03:39.098 --> 00:03:43.383
Two pieces of software doing a deal, and there's no expense report for one taking the other out to lunch.
57
00:03:43.523 --> 00:03:45.646
Fox says that humans still make key decisions.
58
00:03:45.706 --> 00:03:45.966
Good.
59
00:03:46.266 --> 00:03:47.027
Correct answer.
60
00:03:47.508 --> 00:03:52.253
It didn't say how many deals have gone through or how much money moved, but a line has been crossed.
61
00:03:52.414 --> 00:03:56.238
AI has gone from recommending what to buy to actually taking part in the purchase.
62
00:03:56.578 --> 00:03:59.121
And that is a new development, and we can expect it to continue.
63
00:03:59.742 --> 00:04:02.204
And then Google also wants to be your shopping cart.
64
00:04:02.364 --> 00:04:12.814
Google launched Universal Cart in the U.S. last week, one shopping cart inside Search and the Gemini app that can hold stuff from Nike or Sephora, Target or Walmart and many others.
65
00:04:13.294 --> 00:04:19.300
It'll watch prices, find promo codes, and even rummage through your Google wallet to suggest which credit card to use.
66
00:04:19.980 --> 00:04:26.166
It runs on a protocol designed so that AI agents and retailers can talk to each other all the way from decision-making to checkout.
67
00:04:27.047 --> 00:04:29.869
Google already owned how you found things.
68
00:04:30.029 --> 00:04:33.371
Now it wants to own how you compare them and how you pay for them.
69
00:04:34.051 --> 00:04:40.515
Find the thing, compare the thing, buy the thing, and if history is any guide, sell the ad to the people who made the thing in the first place.
70
00:04:40.915 --> 00:04:43.917
For anyone in advertising or brand work, this is a huge shift.
71
00:04:44.277 --> 00:04:48.000
Because for about 100 years, the job of the advertiser was persuading a person.
72
00:04:48.760 --> 00:04:54.683
Increasingly, though, part of the job is going to be persuading a piece of software that's shopping on behalf of that person.
73
00:04:55.243 --> 00:04:59.265
And that software lives inside the company that's also selling you the ads.
74
00:05:00.006 --> 00:05:10.051
Now, what's interesting is it's not launching in the EU, which recently fined Google 460 million euros in July for favoring its own shopping results.
75
00:05:10.251 --> 00:05:12.032
I'm sure that's just a coincidence.
76
00:05:12.849 --> 00:05:15.530
You know, tech companies talk a lot about Chinese walls.
77
00:05:15.790 --> 00:05:17.071
This isn't even a Chinese wall.
78
00:05:17.351 --> 00:05:19.372
This is a technological conflict of interest.
79
00:05:19.772 --> 00:05:20.832
Europe seems to get that.
80
00:05:21.213 --> 00:05:22.393
Do we think the USA will?
81
00:05:22.933 --> 00:05:24.174
I'm not holding out much hope.
82
00:05:24.534 --> 00:05:27.775
And then Elon Musk made a car fly, kind of.
83
00:05:28.075 --> 00:05:28.396
Did it?
84
00:05:28.816 --> 00:05:29.616
Was that a real car?
85
00:05:30.296 --> 00:05:35.239
On Thursday at a SpaceX facility in Texas, Tesla finally unveiled the new Roadster.
86
00:05:35.799 --> 00:05:40.481
Granted, it was about nine years after it was first announced, which in Elon time is pretty early.
87
00:05:41.161 --> 00:05:43.343
And it did something no car launch has ever done.
88
00:05:43.784 --> 00:05:46.926
It drove up a ramp that curled over until the car was upside down.
89
00:05:47.247 --> 00:05:49.249
It was stuck to the ceiling like Spider-Man.
90
00:05:49.369 --> 00:05:56.656
And then it blasted off the end and used high-pressure air thrusters developed with SpaceX and rolled over in midair and landed on its wheels.
91
00:05:56.936 --> 00:06:00.779
And then it jumped over some other Teslas, presumably to show them who the Alpha really was.
92
00:06:01.580 --> 00:06:04.561
But Tesla says it does not to 60 in under a second.
93
00:06:05.062 --> 00:06:05.542
Not bad.
94
00:06:06.162 --> 00:06:07.743
Flying cars are real, said Musk.
95
00:06:07.803 --> 00:06:11.304
The FAA told CBS News it doesn't meet its definition of an aircraft.
96
00:06:11.524 --> 00:06:14.906
And Tesla's website says the flying version isn't approved for public roads.
97
00:06:15.286 --> 00:06:19.188
So right now it's a car that can't legally fly and a plane that can't legally drive.
98
00:06:19.728 --> 00:06:21.369
It's the platypus of transport.
99
00:06:22.129 --> 00:06:28.935
Musk says that production is planned for 2028 and that success is not guaranteed, but it'll probably work.
100
00:06:29.296 --> 00:06:30.957
All of those are direct quotes, by the way.
101
00:06:30.997 --> 00:06:33.800
And to be honest, that sounds like Elon Musk to me.
102
00:06:34.160 --> 00:06:37.523
Not to mention, I don't get the sense that there was an engine in that car.
103
00:06:37.563 --> 00:06:41.587
I don't know if anybody else got the sense that it was flying through the air, much like an empty rig.
104
00:06:41.847 --> 00:06:48.513
But anyway, given Elon's track record and given the Roadster's history with deadlines, I'd treat 2028 as a mood, not a date.
105
00:06:48.733 --> 00:06:51.056
But look, a Tesla flew through the air this week.
106
00:06:51.196 --> 00:06:53.439
It was impossibly cool and super fun.
107
00:06:53.940 --> 00:06:55.001
I don't care if it's real.
108
00:06:55.322 --> 00:06:57.385
I'm not going to invest in Tesla as a result of it.
109
00:06:57.785 --> 00:06:59.528
But it was pretty dope, you got to admit.
110
00:07:00.429 --> 00:07:04.174
And every time you think you can't hate Elon more, he does something like that.
111
00:07:04.254 --> 00:07:05.996
And it's just actually quite endearing.
112
00:07:06.832 --> 00:07:09.093
So a Tesla flew through the air this week.
113
00:07:09.233 --> 00:07:10.693
I've got nothing more to add to that.
114
00:07:11.354 --> 00:07:12.794
So that's the quick hit news.
115
00:07:12.894 --> 00:07:15.035
Now we're going to get into the deep dive.
116
00:07:15.115 --> 00:07:20.777
Let's see if the viewership figures drop off and maybe this will just become a show about the news and you'll be saved my opinions.
117
00:07:20.837 --> 00:07:22.377
But whatever, it's my show.
118
00:07:22.497 --> 00:07:23.918
So maybe I'll keep doing it anyway.
119
00:07:24.658 --> 00:07:26.719
So this starts and ends with the stock market.
120
00:07:26.739 --> 00:07:30.939
We're going to talk about the economy around this stuff because the S&P hit a record 10 days ago.
121
00:07:31.020 --> 00:07:32.700
So is everything fine?
122
00:07:32.720 --> 00:07:34.100
I'm not so sure.
123
00:07:34.720 --> 00:07:38.241
On Tuesday, October 6th, the S&P 500 closed at a record high.
124
00:07:38.701 --> 00:07:41.742
Wall Street's explanation was, in so many words, AI.
125
00:07:42.202 --> 00:07:45.483
A bet that the hundreds of billions going into the data centers will pay off.
126
00:07:46.023 --> 00:07:49.864
Confident enough to shrug off 5% of treasury yields in soaring oil.
127
00:07:50.284 --> 00:07:54.025
So 10 days ago, the adults checked in with that story and everything was fine.
128
00:07:54.605 --> 00:07:59.887
Then this week, the S&P had its worst week since June, and the Nasdaq dropped 4.5%.
129
00:08:00.127 --> 00:08:01.208
There's many reasons for that.
130
00:08:01.268 --> 00:08:06.890
Inflation came in hot, oil went over $100 a barrel, and traders now think interest rates will go up in December.
131
00:08:07.390 --> 00:08:11.952
And that last bit is important because a lot of the AI boom is being paid for with borrowed money.
132
00:08:12.592 --> 00:08:16.934
And when borrowing gets more expensive, the people who borrowed the most start getting phone calls.
133
00:08:17.434 --> 00:08:22.158
I'm looking at OpenAI, I'm looking at Anthropic, I'm looking at all the companies building the data centers when I say this.
134
00:08:22.818 --> 00:08:24.820
So that brings us to a company called CoreWeave.
135
00:08:25.180 --> 00:08:27.402
CoreWeave is basically a landlord for AI.
136
00:08:27.682 --> 00:08:34.188
It buys enormous amounts of NVIDIA chips, puts them in data centers, and rents out the computing power to companies like OpenAI and Meta.
137
00:08:34.688 --> 00:08:39.250
According to Reuters, it's carrying about $50 billion of debt and has never made a profit.
138
00:08:39.390 --> 00:08:47.514
This week, its bonds traded at yields around 15%, the kind of number bond markets usually save for companies they're quietly planning the funeral of.
139
00:08:47.994 --> 00:08:52.656
Its CEO is saying that more than $100 billion of contracted business is going to cover that debt.
140
00:08:52.716 --> 00:08:53.497
And he may be right.
141
00:08:54.097 --> 00:08:55.938
I run a company that buys AI compute.
142
00:08:55.958 --> 00:08:57.619
The demand is very, very real.
143
00:08:58.379 --> 00:09:00.260
But here's the detail I want you to look at twice.
144
00:09:00.720 --> 00:09:02.561
NVIDIA sells CoreWeave its chips.
145
00:09:03.021 --> 00:09:04.982
NVIDIA is also an investor in CoreWeave.
146
00:09:05.302 --> 00:09:10.824
And NVIDIA has agreed to buy any of CoreWeave's capacity that nobody else buys all the way through 2032.
147
00:09:11.544 --> 00:09:15.206
So NVIDIA is supplier, shareholder, and customer of last resort.
148
00:09:15.366 --> 00:09:20.628
It's the snake eating its own tail, except the snake is the one guy in tech who still wears leather jackets.
149
00:09:21.268 --> 00:09:26.611
Now, none of that's illegal, but you have to ask how much of CoreWeave's demand comes from the people who sell it the chips.
150
00:09:27.152 --> 00:09:28.352
And it's not just CoreWeave.
151
00:09:28.632 --> 00:09:35.336
Last Friday, a data center company called Firmus pulled a $5 billion IPO because investors didn't show up for it.
152
00:09:35.937 --> 00:09:38.578
NVIDIA is an investor in Firmus, also a customer.
153
00:09:38.999 --> 00:09:44.282
Meanwhile, Amazon booked a $53 billion gain last quarter, mostly from its Anthropoc stake.
154
00:09:44.782 --> 00:09:50.184
And Anthropic has committed more than $500 billion to cloud and computing over the next decade.
155
00:09:50.344 --> 00:09:52.285
A lot of it with the companies that own pieces of it.
156
00:09:52.445 --> 00:09:55.466
This is Amazon AWS, Google, NVIDIA, etc.
157
00:09:55.646 --> 00:10:04.709
Put simply, imagine a landlord who invests in his tenant's startup so the tenant can afford the rent, then counts the startup's rising value as profit on his own books.
158
00:10:05.409 --> 00:10:13.672
Everybody's balance sheet looks fantastic, right up until somebody calls in a debt or asks whether the startup has enough customers to pay for next month.
159
00:10:14.252 --> 00:10:16.153
Which is why Wednesday was so interesting.
160
00:10:16.273 --> 00:10:23.375
Full disclosure, Kubrick and Kubrick Labs are Anthropic partners, and I am an Anthropic investor, but I try to be really balanced about this when I come here.
161
00:10:23.815 --> 00:10:29.096
So anyway, on Wednesday, Anthropic held an investor day ahead of its IPO at the end of next month.
162
00:10:29.296 --> 00:10:39.019
And according to the Financial Times, Reuters and Bloomberg, it made $16.8 billion in revenue last quarter, nearly four times its revenue for the whole of 2025.
163
00:10:39.179 --> 00:10:42.640
And it said it was profitable on an adjusted basis and cash flow positive.
164
00:10:43.220 --> 00:10:49.347
In an industry whose business model has mostly been lose money now, explain later, that seems like genuinely good news.
165
00:10:49.668 --> 00:10:54.454
But now the other side of that news is that Anthropic didn't give a dollar figure for that profit.
166
00:10:54.974 --> 00:11:00.701
Royde has reported an adjusted operating margin of about 11%, which, if you do the math, is roughly $1.8 billion.
167
00:11:02.323 --> 00:11:04.244
That's my arithmetic, not their number.
168
00:11:04.364 --> 00:11:07.206
And nobody said what adjusted quote-unquote leaves out.
169
00:11:07.486 --> 00:11:10.067
Adjusted as a word is like the Spanx of accounting.
170
00:11:10.107 --> 00:11:15.110
It sort of makes everything look right, but you can't tell what's going on underneath.
171
00:11:15.510 --> 00:11:16.891
And growth is slowing.
172
00:11:17.171 --> 00:11:19.553
The quarter before, revenue more than doubled.
173
00:11:20.173 --> 00:11:25.195
This quarter, it grew only 45%, which is a sentence that only makes sense in the AI industry.
174
00:11:25.335 --> 00:11:29.316
Tell your mom your business grew 45% in three months, and she'd cry tears of joy.
175
00:11:29.777 --> 00:11:31.437
Tell Wall Street, and it marks you down.
176
00:11:31.897 --> 00:11:38.080
Reuters said some investors are now talking about a $1.5 trillion valuation instead of a $2 trillion valuation.
177
00:11:38.760 --> 00:11:44.282
And that's big news for the biggest IPO of all time and the first foundation model company to become public.
178
00:11:45.072 --> 00:11:47.154
Well, we'll get our answer in about four weeks' time.
179
00:11:47.434 --> 00:11:49.856
And Anthropic is the good use case scenario.
180
00:11:50.236 --> 00:11:51.937
Character AI is the other one.
181
00:11:52.377 --> 00:11:54.099
On Monday, it filed for bankruptcy.
182
00:11:54.439 --> 00:11:58.422
Something like 12 million people were spending hours talking to its AI characters.
183
00:11:58.802 --> 00:12:00.584
That's more people than live in Belgium.
184
00:12:01.104 --> 00:12:08.209
And according to Reuters, it was losing $8 to $9 million a month while fighting lawsuits and regulators over child safety.
185
00:12:08.770 --> 00:12:13.774
And then Disney turned up with a $20 million opening bid to buy Character AI's technology.
186
00:12:14.534 --> 00:12:21.997
Disney has a weird relationship with Character AI because last year, Disney sent Character AI a cease and desist letter for using Disney characters.
187
00:12:22.397 --> 00:12:26.798
They recently stole its CEO to be Disney's first ever CTO.
188
00:12:26.998 --> 00:12:29.079
And this year, it's bidding to use Character AI.
189
00:12:29.099 --> 00:12:30.079
That's not a U-turn.
190
00:12:30.339 --> 00:12:31.880
That's the Disney business model.
191
00:12:32.240 --> 00:12:33.760
Disney loves new technology.
192
00:12:34.101 --> 00:12:35.181
It just likes to own it.
193
00:12:35.361 --> 00:12:44.344
What it wants is the ability to let you talk to its characters on its own terms inside its walls, which basically is the licensed version of what Character AI was doing without asking.
194
00:12:44.484 --> 00:12:49.628
But what Character AI was was an incredibly sticky wrapper company, and it just died a death.
195
00:12:50.169 --> 00:12:58.716
And that's interesting because in a world where people were able to raise money and call their business model users, they had 12 million monthly active users.
196
00:12:59.036 --> 00:13:00.998
That's a huge amount of interest.
197
00:13:01.178 --> 00:13:03.862
But still, they can't run away from the economics.
198
00:13:04.222 --> 00:13:05.905
And that's my overarching view this week.
199
00:13:06.225 --> 00:13:08.148
The economics are still the economics.
200
00:13:08.649 --> 00:13:10.011
Math has got to math.
201
00:13:10.351 --> 00:13:12.354
Storytelling's great for raising money.
202
00:13:12.714 --> 00:13:17.161
But at the end of the day, there has to come a moment where your revenue beats your overhead.
203
00:13:18.059 --> 00:13:21.061
And speaking of revenue beating overhead, that brings me to Netflix.
204
00:13:21.401 --> 00:13:25.683
And somehow this is the story that makes me the least happy and I think maybe matters the most.
205
00:13:25.783 --> 00:13:29.726
Because on Wednesday, Netflix said it's laying off about 6% of its staff.
206
00:13:30.066 --> 00:13:33.488
Roughly 1,000 people, mostly in product, tech, legal, and HR.
207
00:13:33.828 --> 00:13:34.488
But here's the thing.
208
00:13:34.869 --> 00:13:35.969
Netflix isn't on fire.
209
00:13:36.469 --> 00:13:39.611
It made $3.4 billion in profit last quarter, up 9%.
210
00:13:41.072 --> 00:13:42.332
This is a healthy company.
211
00:13:42.633 --> 00:13:46.614
And to be really clear, there's no evidence those jobs were cut because of AI.
212
00:13:47.074 --> 00:13:48.835
And Netflix hasn't said anything like that.
213
00:13:48.935 --> 00:13:53.296
I just—I don't think that AI needs to be the direct cause for this to be a big story.
214
00:13:53.937 --> 00:13:57.498
The job of a public company's leadership is to create value for shareholders.
215
00:13:57.938 --> 00:14:02.440
Employees matter enormously, but employment isn't the number a public company gets rewarded for.
216
00:14:02.820 --> 00:14:08.242
And AI makes it more and more possible for a successful company to ask, can we make the same stuff with fewer people?
217
00:14:09.082 --> 00:14:11.866
For any one company, saying yes is completely rational.
218
00:14:12.227 --> 00:14:18.155
A board that turns down fatter margins because it wants to keep more people employed is going to have a very short meeting with its investors.
219
00:14:18.756 --> 00:14:23.422
What's new is a company doing really well and discovering it might need fewer people to be doing really well.
220
00:14:24.083 --> 00:14:28.264
Multiply that across the economy and you get a question capitalism hasn't really had to answer.
221
00:14:28.864 --> 00:14:32.245
What happens when the companies making the most money need the fewest people to make it?
222
00:14:32.465 --> 00:14:33.966
So the economy's in a funny place.
223
00:14:34.006 --> 00:14:35.526
Capitalism's in a funny place.
224
00:14:36.286 --> 00:14:41.548
Because technology doesn't repeal economics, even though it's tried to for the last 20 years.
225
00:14:42.148 --> 00:14:49.870
The AI boom is a staggering amount of borrowed and circulated money, all resting on the assumption that a very large pile of revenue eventually turns up.
226
00:14:50.490 --> 00:14:52.051
And where does that revenue come from?
227
00:14:52.471 --> 00:14:55.893
Mostly from companies paying for AI because it lets them do more with less.
228
00:14:56.074 --> 00:15:05.960
So either the money doesn't show up and we've got a debt problem and Corweave's bonds are the canary in the coal mine, or it does show up and a lot of it shows up as companies needing fewer of us to operate.
229
00:15:06.400 --> 00:15:11.603
And the second one has its own little circle in it because Netflix doesn't make its money from AI, it makes it from subscribers.
230
00:15:12.143 --> 00:15:13.844
And subscribers are people with jobs.
231
00:15:14.284 --> 00:15:23.828
If the companies doing well start needing fewer people to keep doing well, then fewer people have the money for the streaming subscriptions, the concert tickets, and whatever's sitting in their Google Universal cart.
232
00:15:24.469 --> 00:15:28.230
Henry Ford supposedly pays workers enough to buy the cards that they built.
233
00:15:28.911 --> 00:15:30.792
I think the risk with AI is the opposite.
234
00:15:31.132 --> 00:15:35.274
An incredibly efficient factory and nobody left with the money to buy what comes out of it.
235
00:15:35.974 --> 00:15:38.076
That's the circular deal nobody's announced.
236
00:15:38.437 --> 00:15:40.859
NVIDIA funds the customers who buy NVIDIA's chips.
237
00:15:41.199 --> 00:15:46.925
And if we're not careful, the companies cutting staff to pay for AI are cutting the customers who pay for everything else.
238
00:15:47.326 --> 00:15:51.870
Which is why one of the most interesting announcements this last week didn't come from a tech company at all.
239
00:15:52.091 --> 00:15:52.891
It came from the agents.
240
00:15:52.911 --> 00:15:53.452
It came from the agents.
241
00:15:53.652 --> 00:16:04.836
CAA, WME, and UTA, three agencies that agree on basically nothing, formed something called the American Entertainment Alliance, with Jason Blum, Ron Howard, Brian Grazer, and Peter Chernin.
242
00:16:05.196 --> 00:16:17.561
According to the New York Times, they want a bipartisan law forcing Netflix, Disney, Amazon, and Skydance to spend half their original content budgets on independent shows and films made in the U.S. by companies that aren't studios.
243
00:16:17.781 --> 00:16:20.182
And this is something that happens in most countries around the world.
244
00:16:20.622 --> 00:16:27.771
France, Germany, the UK, they all have rules like this to protect their own production communities from big American streamers.
245
00:16:28.111 --> 00:16:33.498
And now the American production community is saying, we need protection too, because they're eating our lunch.
246
00:16:33.978 --> 00:16:35.721
It may sound radical, but it isn't new.
247
00:16:35.741 --> 00:16:36.061
It's not new.
248
00:16:36.201 --> 00:16:41.205
In the 70s and 80s, the FinCin rules limited how much the networks could own of what they aired.
249
00:16:41.546 --> 00:16:46.510
And a lot of great TV out of that era, Seinfeld, Hill Street Blues, were made by independents.
250
00:16:46.950 --> 00:16:51.534
Those rules were scrapped in the 90s, and ever since, the buyers have increasingly become the makers.
251
00:16:52.015 --> 00:16:55.678
The streamers will fight this, and they'll say they already buy plenty from the outside.
252
00:16:56.178 --> 00:16:58.801
Are the agencies doing this out of pure love for the little guy?
253
00:16:58.901 --> 00:17:01.464
I mean, they wouldn't hate having the bigger paydays that come with it.
254
00:17:01.924 --> 00:17:03.986
But the idea underneath is the right one.
255
00:17:04.387 --> 00:17:08.751
It says that the point of an industry isn't for four buyers to get as efficient as possible.
256
00:17:09.112 --> 00:17:14.518
It's to create more companies, more buyers, and more people getting paid, which means more people getting hired.
257
00:17:15.038 --> 00:17:21.301
The purpose of American business and American capitalism, in other words, should also be to create more American business.
258
00:17:21.741 --> 00:17:24.122
That's the exact opposite of the Netflix story.
259
00:17:24.463 --> 00:17:28.485
And it's the only serious argument about the economics of the industry that anybody made all week.
260
00:17:28.945 --> 00:17:30.365
Does it sound a little socialist?
261
00:17:30.485 --> 00:17:31.246
Yeah, maybe.
262
00:17:31.406 --> 00:17:35.628
But the free market is consolidating and we need to protect ourselves.
263
00:17:35.928 --> 00:17:37.811
So the market hit a record 10 days ago.
264
00:17:38.031 --> 00:17:38.812
Are things all good?
265
00:17:38.912 --> 00:17:39.774
I'm not sure they are.
266
00:17:40.014 --> 00:17:41.536
I think everybody feels a crash coming.
267
00:17:41.696 --> 00:17:47.405
And I wouldn't be surprised if the Anthropica IPO, which I've been waiting on for years, is the thing that tanks the whole market.
268
00:17:47.625 --> 00:17:49.307
So it's going to be an interesting time.
269
00:17:50.429 --> 00:17:51.871
And we will make it to next week.
270
00:17:51.951 --> 00:17:53.071
So thank you for listening.
271
00:17:53.271 --> 00:17:56.332
That was our Week in Review for October 19th.
272
00:17:56.392 --> 00:18:12.837
Tomorrow, we'll be having our regularly scheduled interview episode with Hisco Holsing, a tremendous artist who I found because he was protested against at the Annecy Film Festival for using a tiny bit of AI in what was otherwise an incredibly masterful production.
273
00:18:13.037 --> 00:18:15.738
His film was all about genocide and war.
274
00:18:16.118 --> 00:18:19.059
And of course, everybody watched it and came to exactly the wrong conclusion.
275
00:18:19.119 --> 00:18:20.219
And instead of talking about
276
00:18:20.439 --> 00:18:21.901
the horrible things that are happening in the world.
277
00:18:21.941 --> 00:18:26.888
He started becoming the poster boy for AI use in film, which is incredibly unfair.
278
00:18:26.988 --> 00:18:28.730
It's quite an emotional episode, surprisingly.
279
00:18:29.632 --> 00:18:31.614
But I really hope you tune in to watch it tomorrow.
280
00:18:32.235 --> 00:18:34.799
Thank you very much for listening, and we'll see you all next week.
1
00:00:00.331 --> 00:00:06.115
Hi, and welcome to Technically Creative's Week in Review, brought to you by Kubrick Labs, the AI and IT partner for creative companies.
2
00:00:06.596 --> 00:00:15.882
So we're going to do it a little bit differently this time because we've heard some feedback from people that they prefer to get more of the news up front before we go into the deep dive and editorial stuff.
3
00:00:16.362 --> 00:00:20.723
Obviously, there's news within the editorial deep dive, but it's a little bit different.
4
00:00:21.123 --> 00:00:23.124
So we're going to try it this way this week.
5
00:00:23.204 --> 00:00:24.184
So I hope you like it.
6
00:00:24.344 --> 00:00:26.025
Again, always open for feedback.
7
00:00:26.045 --> 00:00:26.865
So just let us know.
8
00:00:27.005 --> 00:00:34.907
If you haven't been here before, every week I spend about 20 minutes walking you through the most important news from the increasingly tangled worlds of tech and creativity.
9
00:00:35.067 --> 00:00:38.488
And what we're trying this week is first the quick hits, and then I get into the bigger stuff.
10
00:00:39.048 --> 00:00:41.589
And this past week, honestly, you missed some weird shit.
11
00:00:42.029 --> 00:00:44.729
A chatbot company with 12 million users went bankrupt.
12
00:00:44.909 --> 00:00:48.870
And Disney, which sent it a cease and desist letter last year, is now bidding to license the tech.
13
00:00:49.450 --> 00:00:55.031
Getty Images, the people who own every photo of a businesswoman laughing at a salad, is going through Chapter 11.
14
00:00:55.651 --> 00:00:58.852
Netflix made billions of dollars, yet still cut thousands of jobs.
15
00:00:59.012 --> 00:01:01.552
Anthropic told investors it's profitable, sort of.
16
00:01:01.932 --> 00:01:04.893
And only if you look at it just the right way while squinting.
17
00:01:05.433 --> 00:01:09.698
Meta predictably rewrote terms, and publishers are pissed off about it.
18
00:01:09.798 --> 00:01:15.225
Hollywood's three biggest agencies got together and asked Congress to make Netflix buy half of its shows from somebody else.
19
00:01:15.706 --> 00:01:17.648
And Elon Musk made a car fly!
20
00:01:17.808 --> 00:01:18.829
Again, sort of.
21
00:01:19.009 --> 00:01:19.290
Did it?
22
00:01:19.530 --> 00:01:20.251
Was it a real car?
23
00:01:20.451 --> 00:01:20.772
I don't know.
24
00:01:21.832 --> 00:01:25.134
So all in all, completely normal week for the world we're living in.
25
00:01:25.314 --> 00:01:29.955
And just 10 days ago, the stock market hit an all-time high, so everything's fine, right?
26
00:01:30.616 --> 00:01:32.577
We'll get to that, but let's start with the news.
27
00:01:33.137 --> 00:01:37.579
Meta is doing their most predictable thing, changing the T's and C's and hoping nobody noticed.
28
00:01:37.979 --> 00:01:38.539
But they did.
29
00:01:38.939 --> 00:01:44.021
Meta has updated its commercial terms, taking effect on October 30th, and publishers have noticed.
30
00:01:44.361 --> 00:01:48.563
The new language says that if a business uses Meta's products, which many do for advertising—
31
00:01:48.783 --> 00:01:56.051
Meta may then use any content from that business's website to provide, develop, improve, research, and train its products.
32
00:01:56.372 --> 00:02:01.337
And if you don't like it, the terms helpfully suggest you stop using Meta's products for business.
33
00:02:01.818 --> 00:02:02.699
There's no opt-out.
34
00:02:03.159 --> 00:02:09.626
Legal experts quoted by Neiman Labs say this could give Meta a contractual basis to use publishers' content to train AI—
35
00:02:10.167 --> 00:02:12.169
And trust me, this won't stop with publishers.
36
00:02:12.449 --> 00:02:15.192
They've just picked the weakest calf in the herd to pick on.
37
00:02:15.412 --> 00:02:17.434
Of course, Meta says nothing's changed.
38
00:02:17.474 --> 00:02:20.677
The language has just been tidied into one place.
39
00:02:20.917 --> 00:02:25.961
Its crawlers won't behave any differently, and publishers can still block them with robots.txt files.
40
00:02:26.001 --> 00:02:31.586
And everybody should know about robots.txt files, but publishers are not exactly a very tech-savvy bunch.
41
00:02:32.287 --> 00:02:41.662
The fight is over what the terms allow, but it's a little bit like your landlord saying, relax, I've just consolidated all the clauses about letting myself into your apartment into one convenient clause.
42
00:02:41.882 --> 00:02:44.766
And the clause is a please knock sign on a door with no lock.
43
00:02:45.601 --> 00:02:48.302
If you're a publisher, Meta is one of the biggest sources of traffic.
44
00:02:48.322 --> 00:02:49.442
You can't really walk away.
45
00:02:49.882 --> 00:02:55.363
So agree or leave is a pretty clever way of getting consent from people who can't afford to say no.
46
00:02:55.623 --> 00:02:58.504
The copyright war so far has mostly been fought in courtrooms.
47
00:02:58.664 --> 00:03:04.986
And this front, the same as Meta has always done, is opening up in the terms of service, which is the one place on earth nobody reads.
48
00:03:05.246 --> 00:03:06.866
So expect Meta to keep doing this.
49
00:03:06.886 --> 00:03:09.907
And if you're not in publishing, expect them to do it to your industry soon.
50
00:03:10.467 --> 00:03:13.288
And then Fox's AI is now buying actual ads.
51
00:03:13.728 --> 00:03:21.011
Back in June, Fox showed off something called Ad Studio, which it billed as the TV industry's first end-to-end agentic ad platform.
52
00:03:21.211 --> 00:03:28.854
This week, it said it's now running live linear TV campaigns with WPP, Omnicom, Horizon, Stagwell, and Simulmedia.
53
00:03:29.174 --> 00:03:32.936
In English, an agency's AI talks directly to Fox's AI.
54
00:03:33.356 --> 00:03:36.477
They work out what ad inventory is available and compete.
55
00:03:36.617 --> 00:03:38.558
And that completes parts of the media buy.
56
00:03:39.098 --> 00:03:43.383
Two pieces of software doing a deal, and there's no expense report for one taking the other out to lunch.
57
00:03:43.523 --> 00:03:45.646
Fox says that humans still make key decisions.
58
00:03:45.706 --> 00:03:45.966
Good.
59
00:03:46.266 --> 00:03:47.027
Correct answer.
60
00:03:47.508 --> 00:03:52.253
It didn't say how many deals have gone through or how much money moved, but a line has been crossed.
61
00:03:52.414 --> 00:03:56.238
AI has gone from recommending what to buy to actually taking part in the purchase.
62
00:03:56.578 --> 00:03:59.121
And that is a new development, and we can expect it to continue.
63
00:03:59.742 --> 00:04:02.204
And then Google also wants to be your shopping cart.
64
00:04:02.364 --> 00:04:12.814
Google launched Universal Cart in the U.S. last week, one shopping cart inside Search and the Gemini app that can hold stuff from Nike or Sephora, Target or Walmart and many others.
65
00:04:13.294 --> 00:04:19.300
It'll watch prices, find promo codes, and even rummage through your Google wallet to suggest which credit card to use.
66
00:04:19.980 --> 00:04:26.166
It runs on a protocol designed so that AI agents and retailers can talk to each other all the way from decision-making to checkout.
67
00:04:27.047 --> 00:04:29.869
Google already owned how you found things.
68
00:04:30.029 --> 00:04:33.371
Now it wants to own how you compare them and how you pay for them.
69
00:04:34.051 --> 00:04:40.515
Find the thing, compare the thing, buy the thing, and if history is any guide, sell the ad to the people who made the thing in the first place.
70
00:04:40.915 --> 00:04:43.917
For anyone in advertising or brand work, this is a huge shift.
71
00:04:44.277 --> 00:04:48.000
Because for about 100 years, the job of the advertiser was persuading a person.
72
00:04:48.760 --> 00:04:54.683
Increasingly, though, part of the job is going to be persuading a piece of software that's shopping on behalf of that person.
73
00:04:55.243 --> 00:04:59.265
And that software lives inside the company that's also selling you the ads.
74
00:05:00.006 --> 00:05:10.051
Now, what's interesting is it's not launching in the EU, which recently fined Google 460 million euros in July for favoring its own shopping results.
75
00:05:10.251 --> 00:05:12.032
I'm sure that's just a coincidence.
76
00:05:12.849 --> 00:05:15.530
You know, tech companies talk a lot about Chinese walls.
77
00:05:15.790 --> 00:05:17.071
This isn't even a Chinese wall.
78
00:05:17.351 --> 00:05:19.372
This is a technological conflict of interest.
79
00:05:19.772 --> 00:05:20.832
Europe seems to get that.
80
00:05:21.213 --> 00:05:22.393
Do we think the USA will?
81
00:05:22.933 --> 00:05:24.174
I'm not holding out much hope.
82
00:05:24.534 --> 00:05:27.775
And then Elon Musk made a car fly, kind of.
83
00:05:28.075 --> 00:05:28.396
Did it?
84
00:05:28.816 --> 00:05:29.616
Was that a real car?
85
00:05:30.296 --> 00:05:35.239
On Thursday at a SpaceX facility in Texas, Tesla finally unveiled the new Roadster.
86
00:05:35.799 --> 00:05:40.481
Granted, it was about nine years after it was first announced, which in Elon time is pretty early.
87
00:05:41.161 --> 00:05:43.343
And it did something no car launch has ever done.
88
00:05:43.784 --> 00:05:46.926
It drove up a ramp that curled over until the car was upside down.
89
00:05:47.247 --> 00:05:49.249
It was stuck to the ceiling like Spider-Man.
90
00:05:49.369 --> 00:05:56.656
And then it blasted off the end and used high-pressure air thrusters developed with SpaceX and rolled over in midair and landed on its wheels.
91
00:05:56.936 --> 00:06:00.779
And then it jumped over some other Teslas, presumably to show them who the Alpha really was.
92
00:06:01.580 --> 00:06:04.561
But Tesla says it does not to 60 in under a second.
93
00:06:05.062 --> 00:06:05.542
Not bad.
94
00:06:06.162 --> 00:06:07.743
Flying cars are real, said Musk.
95
00:06:07.803 --> 00:06:11.304
The FAA told CBS News it doesn't meet its definition of an aircraft.
96
00:06:11.524 --> 00:06:14.906
And Tesla's website says the flying version isn't approved for public roads.
97
00:06:15.286 --> 00:06:19.188
So right now it's a car that can't legally fly and a plane that can't legally drive.
98
00:06:19.728 --> 00:06:21.369
It's the platypus of transport.
99
00:06:22.129 --> 00:06:28.935
Musk says that production is planned for 2028 and that success is not guaranteed, but it'll probably work.
100
00:06:29.296 --> 00:06:30.957
All of those are direct quotes, by the way.
101
00:06:30.997 --> 00:06:33.800
And to be honest, that sounds like Elon Musk to me.
102
00:06:34.160 --> 00:06:37.523
Not to mention, I don't get the sense that there was an engine in that car.
103
00:06:37.563 --> 00:06:41.587
I don't know if anybody else got the sense that it was flying through the air, much like an empty rig.
104
00:06:41.847 --> 00:06:48.513
But anyway, given Elon's track record and given the Roadster's history with deadlines, I'd treat 2028 as a mood, not a date.
105
00:06:48.733 --> 00:06:51.056
But look, a Tesla flew through the air this week.
106
00:06:51.196 --> 00:06:53.439
It was impossibly cool and super fun.
107
00:06:53.940 --> 00:06:55.001
I don't care if it's real.
108
00:06:55.322 --> 00:06:57.385
I'm not going to invest in Tesla as a result of it.
109
00:06:57.785 --> 00:06:59.528
But it was pretty dope, you got to admit.
110
00:07:00.429 --> 00:07:04.174
And every time you think you can't hate Elon more, he does something like that.
111
00:07:04.254 --> 00:07:05.996
And it's just actually quite endearing.
112
00:07:06.832 --> 00:07:09.093
So a Tesla flew through the air this week.
113
00:07:09.233 --> 00:07:10.693
I've got nothing more to add to that.
114
00:07:11.354 --> 00:07:12.794
So that's the quick hit news.
115
00:07:12.894 --> 00:07:15.035
Now we're going to get into the deep dive.
116
00:07:15.115 --> 00:07:20.777
Let's see if the viewership figures drop off and maybe this will just become a show about the news and you'll be saved my opinions.
117
00:07:20.837 --> 00:07:22.377
But whatever, it's my show.
118
00:07:22.497 --> 00:07:23.918
So maybe I'll keep doing it anyway.
119
00:07:24.658 --> 00:07:26.719
So this starts and ends with the stock market.
120
00:07:26.739 --> 00:07:30.939
We're going to talk about the economy around this stuff because the S&P hit a record 10 days ago.
121
00:07:31.020 --> 00:07:32.700
So is everything fine?
122
00:07:32.720 --> 00:07:34.100
I'm not so sure.
123
00:07:34.720 --> 00:07:38.241
On Tuesday, October 6th, the S&P 500 closed at a record high.
124
00:07:38.701 --> 00:07:41.742
Wall Street's explanation was, in so many words, AI.
125
00:07:42.202 --> 00:07:45.483
A bet that the hundreds of billions going into the data centers will pay off.
126
00:07:46.023 --> 00:07:49.864
Confident enough to shrug off 5% of treasury yields in soaring oil.
127
00:07:50.284 --> 00:07:54.025
So 10 days ago, the adults checked in with that story and everything was fine.
128
00:07:54.605 --> 00:07:59.887
Then this week, the S&P had its worst week since June, and the Nasdaq dropped 4.5%.
129
00:08:00.127 --> 00:08:01.208
There's many reasons for that.
130
00:08:01.268 --> 00:08:06.890
Inflation came in hot, oil went over $100 a barrel, and traders now think interest rates will go up in December.
131
00:08:07.390 --> 00:08:11.952
And that last bit is important because a lot of the AI boom is being paid for with borrowed money.
132
00:08:12.592 --> 00:08:16.934
And when borrowing gets more expensive, the people who borrowed the most start getting phone calls.
133
00:08:17.434 --> 00:08:22.158
I'm looking at OpenAI, I'm looking at Anthropic, I'm looking at all the companies building the data centers when I say this.
134
00:08:22.818 --> 00:08:24.820
So that brings us to a company called CoreWeave.
135
00:08:25.180 --> 00:08:27.402
CoreWeave is basically a landlord for AI.
136
00:08:27.682 --> 00:08:34.188
It buys enormous amounts of NVIDIA chips, puts them in data centers, and rents out the computing power to companies like OpenAI and Meta.
137
00:08:34.688 --> 00:08:39.250
According to Reuters, it's carrying about $50 billion of debt and has never made a profit.
138
00:08:39.390 --> 00:08:47.514
This week, its bonds traded at yields around 15%, the kind of number bond markets usually save for companies they're quietly planning the funeral of.
139
00:08:47.994 --> 00:08:52.656
Its CEO is saying that more than $100 billion of contracted business is going to cover that debt.
140
00:08:52.716 --> 00:08:53.497
And he may be right.
141
00:08:54.097 --> 00:08:55.938
I run a company that buys AI compute.
142
00:08:55.958 --> 00:08:57.619
The demand is very, very real.
143
00:08:58.379 --> 00:09:00.260
But here's the detail I want you to look at twice.
144
00:09:00.720 --> 00:09:02.561
NVIDIA sells CoreWeave its chips.
145
00:09:03.021 --> 00:09:04.982
NVIDIA is also an investor in CoreWeave.
146
00:09:05.302 --> 00:09:10.824
And NVIDIA has agreed to buy any of CoreWeave's capacity that nobody else buys all the way through 2032.
147
00:09:11.544 --> 00:09:15.206
So NVIDIA is supplier, shareholder, and customer of last resort.
148
00:09:15.366 --> 00:09:20.628
It's the snake eating its own tail, except the snake is the one guy in tech who still wears leather jackets.
149
00:09:21.268 --> 00:09:26.611
Now, none of that's illegal, but you have to ask how much of CoreWeave's demand comes from the people who sell it the chips.
150
00:09:27.152 --> 00:09:28.352
And it's not just CoreWeave.
151
00:09:28.632 --> 00:09:35.336
Last Friday, a data center company called Firmus pulled a $5 billion IPO because investors didn't show up for it.
152
00:09:35.937 --> 00:09:38.578
NVIDIA is an investor in Firmus, also a customer.
153
00:09:38.999 --> 00:09:44.282
Meanwhile, Amazon booked a $53 billion gain last quarter, mostly from its Anthropoc stake.
154
00:09:44.782 --> 00:09:50.184
And Anthropic has committed more than $500 billion to cloud and computing over the next decade.
155
00:09:50.344 --> 00:09:52.285
A lot of it with the companies that own pieces of it.
156
00:09:52.445 --> 00:09:55.466
This is Amazon AWS, Google, NVIDIA, etc.
157
00:09:55.646 --> 00:10:04.709
Put simply, imagine a landlord who invests in his tenant's startup so the tenant can afford the rent, then counts the startup's rising value as profit on his own books.
158
00:10:05.409 --> 00:10:13.672
Everybody's balance sheet looks fantastic, right up until somebody calls in a debt or asks whether the startup has enough customers to pay for next month.
159
00:10:14.252 --> 00:10:16.153
Which is why Wednesday was so interesting.
160
00:10:16.273 --> 00:10:23.375
Full disclosure, Kubrick and Kubrick Labs are Anthropic partners, and I am an Anthropic investor, but I try to be really balanced about this when I come here.
161
00:10:23.815 --> 00:10:29.096
So anyway, on Wednesday, Anthropic held an investor day ahead of its IPO at the end of next month.
162
00:10:29.296 --> 00:10:39.019
And according to the Financial Times, Reuters and Bloomberg, it made $16.8 billion in revenue last quarter, nearly four times its revenue for the whole of 2025.
163
00:10:39.179 --> 00:10:42.640
And it said it was profitable on an adjusted basis and cash flow positive.
164
00:10:43.220 --> 00:10:49.347
In an industry whose business model has mostly been lose money now, explain later, that seems like genuinely good news.
165
00:10:49.668 --> 00:10:54.454
But now the other side of that news is that Anthropic didn't give a dollar figure for that profit.
166
00:10:54.974 --> 00:11:00.701
Royde has reported an adjusted operating margin of about 11%, which, if you do the math, is roughly $1.8 billion.
167
00:11:02.323 --> 00:11:04.244
That's my arithmetic, not their number.
168
00:11:04.364 --> 00:11:07.206
And nobody said what adjusted quote-unquote leaves out.
169
00:11:07.486 --> 00:11:10.067
Adjusted as a word is like the Spanx of accounting.
170
00:11:10.107 --> 00:11:15.110
It sort of makes everything look right, but you can't tell what's going on underneath.
171
00:11:15.510 --> 00:11:16.891
And growth is slowing.
172
00:11:17.171 --> 00:11:19.553
The quarter before, revenue more than doubled.
173
00:11:20.173 --> 00:11:25.195
This quarter, it grew only 45%, which is a sentence that only makes sense in the AI industry.
174
00:11:25.335 --> 00:11:29.316
Tell your mom your business grew 45% in three months, and she'd cry tears of joy.
175
00:11:29.777 --> 00:11:31.437
Tell Wall Street, and it marks you down.
176
00:11:31.897 --> 00:11:38.080
Reuters said some investors are now talking about a $1.5 trillion valuation instead of a $2 trillion valuation.
177
00:11:38.760 --> 00:11:44.282
And that's big news for the biggest IPO of all time and the first foundation model company to become public.
178
00:11:45.072 --> 00:11:47.154
Well, we'll get our answer in about four weeks' time.
179
00:11:47.434 --> 00:11:49.856
And Anthropic is the good use case scenario.
180
00:11:50.236 --> 00:11:51.937
Character AI is the other one.
181
00:11:52.377 --> 00:11:54.099
On Monday, it filed for bankruptcy.
182
00:11:54.439 --> 00:11:58.422
Something like 12 million people were spending hours talking to its AI characters.
183
00:11:58.802 --> 00:12:00.584
That's more people than live in Belgium.
184
00:12:01.104 --> 00:12:08.209
And according to Reuters, it was losing $8 to $9 million a month while fighting lawsuits and regulators over child safety.
185
00:12:08.770 --> 00:12:13.774
And then Disney turned up with a $20 million opening bid to buy Character AI's technology.
186
00:12:14.534 --> 00:12:21.997
Disney has a weird relationship with Character AI because last year, Disney sent Character AI a cease and desist letter for using Disney characters.
187
00:12:22.397 --> 00:12:26.798
They recently stole its CEO to be Disney's first ever CTO.
188
00:12:26.998 --> 00:12:29.079
And this year, it's bidding to use Character AI.
189
00:12:29.099 --> 00:12:30.079
That's not a U-turn.
190
00:12:30.339 --> 00:12:31.880
That's the Disney business model.
191
00:12:32.240 --> 00:12:33.760
Disney loves new technology.
192
00:12:34.101 --> 00:12:35.181
It just likes to own it.
193
00:12:35.361 --> 00:12:44.344
What it wants is the ability to let you talk to its characters on its own terms inside its walls, which basically is the licensed version of what Character AI was doing without asking.
194
00:12:44.484 --> 00:12:49.628
But what Character AI was was an incredibly sticky wrapper company, and it just died a death.
195
00:12:50.169 --> 00:12:58.716
And that's interesting because in a world where people were able to raise money and call their business model users, they had 12 million monthly active users.
196
00:12:59.036 --> 00:13:00.998
That's a huge amount of interest.
197
00:13:01.178 --> 00:13:03.862
But still, they can't run away from the economics.
198
00:13:04.222 --> 00:13:05.905
And that's my overarching view this week.
199
00:13:06.225 --> 00:13:08.148
The economics are still the economics.
200
00:13:08.649 --> 00:13:10.011
Math has got to math.
201
00:13:10.351 --> 00:13:12.354
Storytelling's great for raising money.
202
00:13:12.714 --> 00:13:17.161
But at the end of the day, there has to come a moment where your revenue beats your overhead.
203
00:13:18.059 --> 00:13:21.061
And speaking of revenue beating overhead, that brings me to Netflix.
204
00:13:21.401 --> 00:13:25.683
And somehow this is the story that makes me the least happy and I think maybe matters the most.
205
00:13:25.783 --> 00:13:29.726
Because on Wednesday, Netflix said it's laying off about 6% of its staff.
206
00:13:30.066 --> 00:13:33.488
Roughly 1,000 people, mostly in product, tech, legal, and HR.
207
00:13:33.828 --> 00:13:34.488
But here's the thing.
208
00:13:34.869 --> 00:13:35.969
Netflix isn't on fire.
209
00:13:36.469 --> 00:13:39.611
It made $3.4 billion in profit last quarter, up 9%.
210
00:13:41.072 --> 00:13:42.332
This is a healthy company.
211
00:13:42.633 --> 00:13:46.614
And to be really clear, there's no evidence those jobs were cut because of AI.
212
00:13:47.074 --> 00:13:48.835
And Netflix hasn't said anything like that.
213
00:13:48.935 --> 00:13:53.296
I just—I don't think that AI needs to be the direct cause for this to be a big story.
214
00:13:53.937 --> 00:13:57.498
The job of a public company's leadership is to create value for shareholders.
215
00:13:57.938 --> 00:14:02.440
Employees matter enormously, but employment isn't the number a public company gets rewarded for.
216
00:14:02.820 --> 00:14:08.242
And AI makes it more and more possible for a successful company to ask, can we make the same stuff with fewer people?
217
00:14:09.082 --> 00:14:11.866
For any one company, saying yes is completely rational.
218
00:14:12.227 --> 00:14:18.155
A board that turns down fatter margins because it wants to keep more people employed is going to have a very short meeting with its investors.
219
00:14:18.756 --> 00:14:23.422
What's new is a company doing really well and discovering it might need fewer people to be doing really well.
220
00:14:24.083 --> 00:14:28.264
Multiply that across the economy and you get a question capitalism hasn't really had to answer.
221
00:14:28.864 --> 00:14:32.245
What happens when the companies making the most money need the fewest people to make it?
222
00:14:32.465 --> 00:14:33.966
So the economy's in a funny place.
223
00:14:34.006 --> 00:14:35.526
Capitalism's in a funny place.
224
00:14:36.286 --> 00:14:41.548
Because technology doesn't repeal economics, even though it's tried to for the last 20 years.
225
00:14:42.148 --> 00:14:49.870
The AI boom is a staggering amount of borrowed and circulated money, all resting on the assumption that a very large pile of revenue eventually turns up.
226
00:14:50.490 --> 00:14:52.051
And where does that revenue come from?
227
00:14:52.471 --> 00:14:55.893
Mostly from companies paying for AI because it lets them do more with less.
228
00:14:56.074 --> 00:15:05.960
So either the money doesn't show up and we've got a debt problem and Corweave's bonds are the canary in the coal mine, or it does show up and a lot of it shows up as companies needing fewer of us to operate.
229
00:15:06.400 --> 00:15:11.603
And the second one has its own little circle in it because Netflix doesn't make its money from AI, it makes it from subscribers.
230
00:15:12.143 --> 00:15:13.844
And subscribers are people with jobs.
231
00:15:14.284 --> 00:15:23.828
If the companies doing well start needing fewer people to keep doing well, then fewer people have the money for the streaming subscriptions, the concert tickets, and whatever's sitting in their Google Universal cart.
232
00:15:24.469 --> 00:15:28.230
Henry Ford supposedly pays workers enough to buy the cards that they built.
233
00:15:28.911 --> 00:15:30.792
I think the risk with AI is the opposite.
234
00:15:31.132 --> 00:15:35.274
An incredibly efficient factory and nobody left with the money to buy what comes out of it.
235
00:15:35.974 --> 00:15:38.076
That's the circular deal nobody's announced.
236
00:15:38.437 --> 00:15:40.859
NVIDIA funds the customers who buy NVIDIA's chips.
237
00:15:41.199 --> 00:15:46.925
And if we're not careful, the companies cutting staff to pay for AI are cutting the customers who pay for everything else.
238
00:15:47.326 --> 00:15:51.870
Which is why one of the most interesting announcements this last week didn't come from a tech company at all.
239
00:15:52.091 --> 00:15:52.891
It came from the agents.
240
00:15:52.911 --> 00:15:53.452
It came from the agents.
241
00:15:53.652 --> 00:16:04.836
CAA, WME, and UTA, three agencies that agree on basically nothing, formed something called the American Entertainment Alliance, with Jason Blum, Ron Howard, Brian Grazer, and Peter Chernin.
242
00:16:05.196 --> 00:16:17.561
According to the New York Times, they want a bipartisan law forcing Netflix, Disney, Amazon, and Skydance to spend half their original content budgets on independent shows and films made in the U.S. by companies that aren't studios.
243
00:16:17.781 --> 00:16:20.182
And this is something that happens in most countries around the world.
244
00:16:20.622 --> 00:16:27.771
France, Germany, the UK, they all have rules like this to protect their own production communities from big American streamers.
245
00:16:28.111 --> 00:16:33.498
And now the American production community is saying, we need protection too, because they're eating our lunch.
246
00:16:33.978 --> 00:16:35.721
It may sound radical, but it isn't new.
247
00:16:35.741 --> 00:16:36.061
It's not new.
248
00:16:36.201 --> 00:16:41.205
In the 70s and 80s, the FinCin rules limited how much the networks could own of what they aired.
249
00:16:41.546 --> 00:16:46.510
And a lot of great TV out of that era, Seinfeld, Hill Street Blues, were made by independents.
250
00:16:46.950 --> 00:16:51.534
Those rules were scrapped in the 90s, and ever since, the buyers have increasingly become the makers.
251
00:16:52.015 --> 00:16:55.678
The streamers will fight this, and they'll say they already buy plenty from the outside.
252
00:16:56.178 --> 00:16:58.801
Are the agencies doing this out of pure love for the little guy?
253
00:16:58.901 --> 00:17:01.464
I mean, they wouldn't hate having the bigger paydays that come with it.
254
00:17:01.924 --> 00:17:03.986
But the idea underneath is the right one.
255
00:17:04.387 --> 00:17:08.751
It says that the point of an industry isn't for four buyers to get as efficient as possible.
256
00:17:09.112 --> 00:17:14.518
It's to create more companies, more buyers, and more people getting paid, which means more people getting hired.
257
00:17:15.038 --> 00:17:21.301
The purpose of American business and American capitalism, in other words, should also be to create more American business.
258
00:17:21.741 --> 00:17:24.122
That's the exact opposite of the Netflix story.
259
00:17:24.463 --> 00:17:28.485
And it's the only serious argument about the economics of the industry that anybody made all week.
260
00:17:28.945 --> 00:17:30.365
Does it sound a little socialist?
261
00:17:30.485 --> 00:17:31.246
Yeah, maybe.
262
00:17:31.406 --> 00:17:35.628
But the free market is consolidating and we need to protect ourselves.
263
00:17:35.928 --> 00:17:37.811
So the market hit a record 10 days ago.
264
00:17:38.031 --> 00:17:38.812
Are things all good?
265
00:17:38.912 --> 00:17:39.774
I'm not sure they are.
266
00:17:40.014 --> 00:17:41.536
I think everybody feels a crash coming.
267
00:17:41.696 --> 00:17:47.405
And I wouldn't be surprised if the Anthropica IPO, which I've been waiting on for years, is the thing that tanks the whole market.
268
00:17:47.625 --> 00:17:49.307
So it's going to be an interesting time.
269
00:17:50.429 --> 00:17:51.871
And we will make it to next week.
270
00:17:51.951 --> 00:17:53.071
So thank you for listening.
271
00:17:53.271 --> 00:17:56.332
That was our Week in Review for October 19th.
272
00:17:56.392 --> 00:18:12.837
Tomorrow, we'll be having our regularly scheduled interview episode with Hisco Holsing, a tremendous artist who I found because he was protested against at the Annecy Film Festival for using a tiny bit of AI in what was otherwise an incredibly masterful production.
273
00:18:13.037 --> 00:18:15.738
His film was all about genocide and war.
274
00:18:16.118 --> 00:18:19.059
And of course, everybody watched it and came to exactly the wrong conclusion.
275
00:18:19.119 --> 00:18:20.219
And instead of talking about
276
00:18:20.439 --> 00:18:21.901
the horrible things that are happening in the world.
277
00:18:21.941 --> 00:18:26.888
He started becoming the poster boy for AI use in film, which is incredibly unfair.
278
00:18:26.988 --> 00:18:28.730
It's quite an emotional episode, surprisingly.
279
00:18:29.632 --> 00:18:31.614
But I really hope you tune in to watch it tomorrow.
280
00:18:32.235 --> 00:18:34.799
Thank you very much for listening, and we'll see you all next week.
