WEBVTT - Monologue: AI's Desperation Era

0:00:02.120 --> 0:00:07.600
<v Speaker 1>Zone Media. Hello on, Welcome to this week's Better Offline Monologue.

0:00:07.640 --> 0:00:18.520
<v Speaker 1>I'm your host ed Zetron. What it's been a peculiar week.

0:00:18.800 --> 0:00:22.079
<v Speaker 1>Two days ago, Sarah Fryar, chief financial officer of open AI,

0:00:22.400 --> 0:00:24.560
<v Speaker 1>said in a stage at a Wall Street Journal conference

0:00:24.600 --> 0:00:27.480
<v Speaker 1>that an IPO isn't on the cards and foolishly said

0:00:27.520 --> 0:00:29.880
<v Speaker 1>the open AI was and I quote looking for an

0:00:29.920 --> 0:00:34.120
<v Speaker 1>ecosystem of banks, private equity, maybe even governmental and that

0:00:34.159 --> 0:00:36.400
<v Speaker 1>the way governments can come to bear can really drop

0:00:36.440 --> 0:00:39.240
<v Speaker 1>the cost of financing but also increase the loan to value,

0:00:39.440 --> 0:00:41.120
<v Speaker 1>so the amount of debt you can take on top

0:00:41.159 --> 0:00:44.880
<v Speaker 1>of an equity portion. Now, she also, according to Bloomberg,

0:00:45.000 --> 0:00:48.320
<v Speaker 1>hinted at some sort of federal backstop, literal government backing

0:00:48.360 --> 0:00:52.280
<v Speaker 1>for loans for data centers. Very strange. Now, I will

0:00:52.280 --> 0:00:55.440
<v Speaker 1>add this as being completely misreported as open AI demanding

0:00:55.440 --> 0:00:57.520
<v Speaker 1>government funds. Had someone ring me the other night say

0:00:57.640 --> 0:00:59.760
<v Speaker 1>did you hear about this? It's not what happened. To

0:00:59.800 --> 0:01:03.240
<v Speaker 1>be clear, I believe absolutely that Sarah Fryer was throwing

0:01:03.280 --> 0:01:05.839
<v Speaker 1>out the idea to see if governments might buy, only

0:01:05.880 --> 0:01:08.120
<v Speaker 1>to find the entirety of social media and most of

0:01:08.160 --> 0:01:12.400
<v Speaker 1>the mainstream media reacting like Friar had kicked the family dog. Nevertheless,

0:01:12.400 --> 0:01:14.080
<v Speaker 1>I want to be clear that open AI has not

0:01:14.160 --> 0:01:17.400
<v Speaker 1>requested a federal backstop, and that as ever, it appears

0:01:17.400 --> 0:01:19.520
<v Speaker 1>that even the big media outlets do not seem to

0:01:19.600 --> 0:01:22.119
<v Speaker 1>go and listen to the things that people say or check.

0:01:22.600 --> 0:01:26.040
<v Speaker 1>It's ridiculous. This is a huge story if you misreported,

0:01:26.360 --> 0:01:30.559
<v Speaker 1>which is what's happened. Nevertheless, as I said on Blue Sky,

0:01:30.720 --> 0:01:33.640
<v Speaker 1>this is an absolute diva moment for open AI. Oh

0:01:33.720 --> 0:01:36.760
<v Speaker 1>I'm so messy. I can't even afford mid data centers. Please,

0:01:36.880 --> 0:01:39.240
<v Speaker 1>Uncle Sam, I promised to spend one point five trillion

0:01:39.280 --> 0:01:41.399
<v Speaker 1>dollars in the next five years. Oh, I'm so weak

0:01:41.440 --> 0:01:43.880
<v Speaker 1>and powerless in the face of all the promises I made.

0:01:44.080 --> 0:01:46.919
<v Speaker 1>Grow up, wipe your own ass. If regular people promised

0:01:46.920 --> 0:01:48.480
<v Speaker 1>a bunch of people a ton of money that they

0:01:48.520 --> 0:01:50.520
<v Speaker 1>couldn't afford, they be sued, but in prison or have

0:01:50.560 --> 0:01:52.880
<v Speaker 1>their assets reduced to zero. Why do we have to

0:01:52.880 --> 0:01:55.840
<v Speaker 1>do something different for open AI? Anyway, this is not

0:01:55.920 --> 0:01:58.559
<v Speaker 1>going to happen. I've never seen a more aggressive bipartisan

0:01:58.600 --> 0:02:01.760
<v Speaker 1>descent for an idea by everyone from better offline listeners

0:02:01.760 --> 0:02:03.920
<v Speaker 1>in the middle of prying hubcaps off of cars to

0:02:03.960 --> 0:02:07.960
<v Speaker 1>the powerful Dunce coalition led by Matty Iglesias. Everybody fucking

0:02:08.000 --> 0:02:10.880
<v Speaker 1>hates this idea, and if I'm honest, I think everybody

0:02:10.880 --> 0:02:12.920
<v Speaker 1>is sick and tired of AI. What was once a

0:02:12.960 --> 0:02:15.400
<v Speaker 1>market wide consensus that AI was the future has now

0:02:15.400 --> 0:02:17.880
<v Speaker 1>become a cacophony of hand ringing in just about every

0:02:17.880 --> 0:02:21.079
<v Speaker 1>major media outlet. What hasn't helped was a very very

0:02:21.120 --> 0:02:23.919
<v Speaker 1>weird post through a Nvidia's social media account, a single

0:02:23.960 --> 0:02:27.560
<v Speaker 1>image that said, statement from Nvidia CEO Jensen Wang. As

0:02:27.600 --> 0:02:30.600
<v Speaker 1>I have long said, China is nanoseconds behind America in AI,

0:02:30.760 --> 0:02:33.360
<v Speaker 1>it's vital that America wins by wrestling ahead and winning

0:02:33.400 --> 0:02:38.760
<v Speaker 1>developers worldwide. Not really sure what that meant, but this

0:02:39.040 --> 0:02:41.000
<v Speaker 1>was I found after the fact, tied to a statement

0:02:41.080 --> 0:02:43.440
<v Speaker 1>Huang made to the Financial Times that mostly said the

0:02:43.440 --> 0:02:45.280
<v Speaker 1>same thing. But I've got to say, this is the

0:02:45.320 --> 0:02:49.359
<v Speaker 1>weirdest statement I've seen. It's so weird. What do you want, Jensen?

0:02:49.480 --> 0:02:54.160
<v Speaker 1>More developers to developers by GPUs or so wait a second,

0:02:54.720 --> 0:02:57.720
<v Speaker 1>China doesn't have the latest in video GPUs. How are

0:02:57.720 --> 0:03:02.200
<v Speaker 1>they nanoseconds behind America? Surely America should be I don't

0:03:02.240 --> 0:03:05.639
<v Speaker 1>know hours days at Jensen, I don't know why you're

0:03:05.680 --> 0:03:08.160
<v Speaker 1>using words that relate to time as measures of distance.

0:03:08.200 --> 0:03:11.720
<v Speaker 1>But nevertheless, why aren't American companies further ahead when they

0:03:11.720 --> 0:03:15.000
<v Speaker 1>have easy access to Blackwell and soon via Rubin GPUs?

0:03:15.360 --> 0:03:18.280
<v Speaker 1>What exactly do you want us to do? Jensen? What

0:03:18.480 --> 0:03:21.040
<v Speaker 1>is the thing? It's like vague posting, being like I

0:03:21.080 --> 0:03:23.680
<v Speaker 1>really shouldn't be alone right now, you know, I just

0:03:23.720 --> 0:03:25.840
<v Speaker 1>I've had such a bad night. I wish someone would

0:03:25.919 --> 0:03:30.480
<v Speaker 1>text me and tell me they're buying a gpu Hanah. No, man,

0:03:30.840 --> 0:03:34.080
<v Speaker 1>you're the CEO of the largest company on the stock market.

0:03:34.400 --> 0:03:37.000
<v Speaker 1>Do you not have a little more bizaz, little more

0:03:37.080 --> 0:03:41.720
<v Speaker 1>riz rizzless? Jensen, You're rizzless even with those beautiful leather jackets.

0:03:42.080 --> 0:03:44.320
<v Speaker 1>And it's almost as if there isn't a coherent reason

0:03:44.400 --> 0:03:46.840
<v Speaker 1>to buy GPUs or any real progress coming out of

0:03:46.880 --> 0:03:50.320
<v Speaker 1>American AI labs, or really any of them were still though,

0:03:50.360 --> 0:03:53.000
<v Speaker 1>and this was the one that really I think was underreported.

0:03:53.280 --> 0:03:56.280
<v Speaker 1>During Microsoft's last earnings, the company reported a four point

0:03:56.360 --> 0:03:58.560
<v Speaker 1>one billion dollar loss on their thirty two point five

0:03:58.800 --> 0:04:01.920
<v Speaker 1>percent staking open air, meaning that due to the company

0:04:02.000 --> 0:04:05.600
<v Speaker 1>using the equity method for accounting OpenAI lost twelve billion

0:04:05.640 --> 0:04:09.000
<v Speaker 1>dollars in the third quarter of this calendar year. Now,

0:04:09.280 --> 0:04:13.120
<v Speaker 1>open Ai CFO Sarah Fryar said that this was way overestimated.

0:04:13.240 --> 0:04:16.799
<v Speaker 1>And I want to be clear about something. Microsoft's earnings

0:04:16.800 --> 0:04:20.239
<v Speaker 1>are public filings as they are a public company, meaning

0:04:20.279 --> 0:04:23.640
<v Speaker 1>that these are looked over by accountants and lawyers. Why

0:04:23.720 --> 0:04:27.760
<v Speaker 1>is the chief financial officer of open ai disputing these numbers.

0:04:28.120 --> 0:04:32.200
<v Speaker 1>That's extremely goddamn weird. And this isn't the first time

0:04:32.240 --> 0:04:34.800
<v Speaker 1>Sarah Friar has said some weird shit, even in this

0:04:34.920 --> 0:04:37.680
<v Speaker 1>conference appearance, but in general. The other time I saw

0:04:37.720 --> 0:04:39.919
<v Speaker 1>her say that it was thirty five dollars a token

0:04:40.279 --> 0:04:43.600
<v Speaker 1>in the past for open AI's models. Thirty five bucks

0:04:43.640 --> 0:04:45.680
<v Speaker 1>per million tokens doesn't even make sense, but I think

0:04:45.680 --> 0:04:47.760
<v Speaker 1>thirty five bucks per token would mean if they spend

0:04:47.760 --> 0:04:50.680
<v Speaker 1>like trillions of dollars a year. Regardless, you're the CFO

0:04:50.720 --> 0:04:52.760
<v Speaker 1>of open Ai. Shouldn't you know this shit a little

0:04:52.800 --> 0:04:55.599
<v Speaker 1>bit better? Now? Before you email me something about gap

0:04:55.640 --> 0:04:58.000
<v Speaker 1>earnings or accounting. I'm going to include links in the

0:04:58.040 --> 0:05:00.520
<v Speaker 1>show notes because I assure you this loss is directly

0:05:00.560 --> 0:05:05.120
<v Speaker 1>reflective of open Eye's actual losses, and that's really weird.

0:05:05.200 --> 0:05:08.359
<v Speaker 1>By the way. It's really weird. Indeed, and now that

0:05:08.400 --> 0:05:11.240
<v Speaker 1>we know that, we have to ask how bad are

0:05:11.320 --> 0:05:15.840
<v Speaker 1>open aye's actual losses? How much worse are things than

0:05:15.880 --> 0:05:20.640
<v Speaker 1>we know? I assure you this loss is real. I

0:05:20.800 --> 0:05:24.560
<v Speaker 1>trust it. It's been well reported, looked over by actual accountants,

0:05:24.600 --> 0:05:26.120
<v Speaker 1>and a link to a Wall Street Journal piece where

0:05:26.120 --> 0:05:28.480
<v Speaker 1>an actual accountant looked at this and made this statement.

0:05:29.279 --> 0:05:31.159
<v Speaker 1>And I got to ask, what is it that Sarah

0:05:31.240 --> 0:05:34.640
<v Speaker 1>Fryar and Clammy Sammy don't want us to know what

0:05:34.839 --> 0:05:38.080
<v Speaker 1>is going on? The Information reported that in the first

0:05:38.120 --> 0:05:40.560
<v Speaker 1>half of this year, open aye made four point three

0:05:40.560 --> 0:05:42.520
<v Speaker 1>billion dollars of revenue but had a net loss of

0:05:42.560 --> 0:05:45.320
<v Speaker 1>thirteen point five billion dollars. But that's in six months.

0:05:46.400 --> 0:05:51.600
<v Speaker 1>Are those indicative of the real numbers? Has someone been fibbing?

0:05:52.080 --> 0:05:54.599
<v Speaker 1>I trust the Information's reporting, by the way, I fully

0:05:54.640 --> 0:05:58.720
<v Speaker 1>believe them, and I at this point am just concerned

0:05:58.839 --> 0:06:02.200
<v Speaker 1>that we're not getting the full picture. I don't know

0:06:02.279 --> 0:06:04.400
<v Speaker 1>what open ai is sharing to people. I don't know

0:06:04.440 --> 0:06:08.320
<v Speaker 1>how these reporters have got these numbers, But that twelve

0:06:08.320 --> 0:06:12.120
<v Speaker 1>billion dollar loss in a quarter that feels quite dramatic,

0:06:12.160 --> 0:06:14.960
<v Speaker 1>and it feels a lot more than open ai has

0:06:14.960 --> 0:06:18.919
<v Speaker 1>been hinting they're losing. What else? Don't we know what

0:06:19.040 --> 0:06:21.800
<v Speaker 1>else is going on in Open Ai that we are

0:06:21.839 --> 0:06:26.520
<v Speaker 1>not Party two? Hopefully we find out soon. In any case,

0:06:26.560 --> 0:06:31.040
<v Speaker 1>it feels like something came unbuckled this week. Glamuel the

0:06:31.040 --> 0:06:42.000
<v Speaker 1>Smiling Man is watching a specter horned Silicon Valley