YouTube2Text

Anthropic-IPO: Ich zeichne! // die große Analyse der Claude-Aktie — Transcript

by Mario Lochner · 6,013 words · 966 segments · language en · Watch on YouTube

Full transcript

  1. 0:001.4 trillion dollars and three letters:
  2. 0:03A-N-T. A company that nobody knew five
  3. 0:06years ago goes public on November 19th.
  4. 0:09November for its IPO and it wants to
  5. 0:11raise more money than SpaceX did in its
  6. 0:14record IPO in June. We are, of course,
  7. 0:16talking about Anthropic, the makers of
  8. 0:18Claude. This is the biggest AI IPO of
  9. 0:20all time, and folks, I’m about to show
  10. 0:22you numbers in the next few minutes that
  11. 0:24you won't even believe yourselves.
  12. 0:29Hello everyone and welcome to a
  13. 0:30brand-new video on my channel. My name
  14. 0:32is Mario Lochner and today we are
  15. 0:34looking at the second major IPO of the
  16. 0:36year.
  17. 0:37It’s about Anthropic, and investors can
  18. 0:38still subscribe until November 18th.
  19. 0:40November. The 19th. November is then the
  20. 0:43first day of trading, and of course, we
  21. 0:44will answer the big question in this
  22. 0:46video.
  23. 0:47Subscribe, buy, or stay away? I have a
  24. 0:51very clear opinion, and if you like my
  25. 0:53videos, you can do me a huge favor and
  26. 0:56subscribe to my channel and, of course,
  27. 0:59leave me a thumbs up if you enjoy it—
  28. 1:01the famous algorithm slap. Let's quickly
  29. 1:03look at Anthropic’s figures before we
  30. 1:05really dive deep into the details. For
  31. 1:07the third quarter, we see 17 billion
  32. 1:09dollars in revenue, which is very strong
  33. 1:11for just three months, and it’s more
  34. 1:12than a multiple of the previous year. We
  35. 1:14see the same trend with customers paying
  36. 1:17more than 100,000 dollars, which now
  37. 1:19number more than 11,900.
  38. 1:21We are seeing massive efficiency gains
  39. 1:23in computing power and also an adjusted
  40. 1:25operating margin of 15%, and now here is
  41. 1:28the number—yeah, it’s really a stunner.
  42. 1:31Because this is the one analysts had to
  43. 1:33look at twice. I also had to look at it
  44. 1:35two or three times because it is truly
  45. 1:36impressive.
  46. 1:37Anyone who spent one million at
  47. 1:40Anthropic a year ago is now spending an
  48. 1:42average of 9 million—from one to nine, a
  49. 1:45net revenue retention of 900%—and that
  50. 1:47will be very, very crucial when we look
  51. 1:50at the valuation and the scenarios later
  52. 1:52.
  53. 1:54So far so good, it sounds like a reason
  54. 1:55to celebrate, and you can be happy now
  55. 1:57too, because Christmas is coming soon
  56. 1:59and Christmas will be better than ever.
  57. 2:01This year there is the Beating Beta
  58. 2:03Advent calendar and you can win every
  59. 2:05day in December. There are prizes worth
  60. 2:08more than 6,000 euros and you can, of
  61. 2:10course, participate for free.
  62. 2:12You can find the link in the video
  63. 2:13description or in the show notes below.
  64. 2:15And the prizes are amazing. We have a
  65. 2:17Montblanc fountain pen, which alone is
  66. 2:19an absolute showstopper.
  67. 2:21But we also have vintage champagne, a
  68. 2:23smartwatch, great books, silver coins,
  69. 2:25gold coins, and much more.
  70. 2:28Check out the prizes, find the link
  71. 2:29below, join for free, and you could win
  72. 2:31every single day in December.
  73. 2:34Now, let's put the joy aside for a
  74. 2:35moment. Of course, there is a flip side
  75. 2:37at Anthropic: a massive net loss of $
  76. 2:39104 billion in the first nine months of
  77. 2:412026.
  78. 2:42It sounds catastrophic at first, but
  79. 2:44there is one thing you need to know. The
  80. 2:45majority of that, a good $ 94 billion,
  81. 2:47is a book loss from convertible bonds,
  82. 2:49so no money is actually flowing out, but
  83. 2:52it is on the balance sheet. And then
  84. 2:53there’s the rescue performance.
  85. 2:55Anthropic has committed to spending
  86. 2:57around $ 900 billion on data centers and
  87. 2:59infrastructure.
  88. 3:01Nearly $ 71 billion of that is due
  89. 3:03within a single year. On top of that
  90. 3:05come other risks and negative headlines
  91. 3:07. A dispute with the Pentagon. The U.S.
  92. 3:10government actually classifies Anthropic
  93. 3:12as a supply chain risk, and the whole
  94. 3:14thing is far from over and remains a
  95. 3:17burden. Big opportunities, big risks.
  96. 3:19How did this whole story come about?
  97. 3:21With a break. Many people might not even
  98. 3:22know this. It was at the end of 2020
  99. 3:25when several senior OpenAI staff left
  100. 3:27their employer. Their accusation: safety
  101. 3:29is being neglected there, so they
  102. 3:31founded Anthropic. The CEO, as is the
  103. 3:34case today, is Dario Amodei. And five
  104. 3:37years later, the outlier Anthropic has
  105. 3:39beaten its rival to the stock market.
  106. 3:41OpenAI wants to go public too, but won't
  107. 3:44manage it before 2027 at the earliest;
  108. 3:46the outcome is open. So, record growth
  109. 3:48on one side, and on the other, we have
  110. 3:50multi-billion dollar risks—and we'll
  111. 3:52take a look at those other risks right
  112. 3:53now. The question isn't whether
  113. 3:56Anthropic is growing; the question is,
  114. 3:58is Anthropic worth $ 1.4 trillion? We’ll
  115. 4:01answer that in a moment. Today we're
  116. 4:03looking at the risks, but also the moat
  117. 4:06, our scenarios for Anthropic, and our
  118. 4:08decision.
  119. 4:09Clearly, we will explain everything
  120. 4:10about the IPO in detail shortly.
  121. 4:12But first, I’d like to introduce an old
  122. 4:14acquaintance who has actually changed
  123. 4:16his mind for Anthropic and is investing
  124. 4:18in individual stocks again for the first
  125. 4:20time in a long while.
  126. 4:21>> Hi everyone, my name is Sinan Krieger.
  127. 4:24I’m a co-founder at Beatvest, and some
  128. 4:27of you might know me from the Locker
  129. 4:29Room Talk and are wondering, "Man, Sinan
  130. 4:32, you've been saying for years that you
  131. 4:35don't invest in individual stocks, but
  132. 4:37only in ETFs."
  133. 4:39And yes, that’s true—or it was until now
  134. 4:42, because I’m making an exception for
  135. 4:46Anthropic. One thing upfront, though,
  136. 4:47which is very important to me.
  137. 4:49My fundamental stance on how I plan and
  138. 4:51structure my long-term wealth
  139. 4:53accumulation has not changed.
  140. 4:56I hold a similar view to Christian Fuchs
  141. 4:58—you likely know him from the interviews
  142. 5:00with Mario—that you should always look
  143. 5:01at your wealth holistically.
  144. 5:04Meaning you shouldn't just look at your
  145. 5:07portfolio, but consciously include all
  146. 5:09areas, because that is the only way to
  147. 5:11accurately reflect your true risk
  148. 5:13profile.
  149. 5:15In my specific case, this means that my
  150. 5:17high-beta satellite, yes, my high-risk
  151. 5:20assets with which I want to beat the
  152. 5:22market long-term and achieve excess
  153. 5:24returns, is my company, my startup
  154. 5:26Beatvest.
  155. 5:28Here I have by far the greatest
  156. 5:30potential, but also by far the greatest
  157. 5:32risk.
  158. 5:33And it is the curse and blessing of a
  159. 5:35typical entrepreneur that, in most cases
  160. 5:38, the greatest wealth lies dormant in
  161. 5:40one's own company.
  162. 5:43That is why my goal has always been to
  163. 5:45remove risk from all other asset areas,
  164. 5:48like the stock portfolio, for example.
  165. 5:52That is also the reason for my "ETF only
  166. 5:54" approach, and why I maintain a
  167. 5:56comparatively very high cash position
  168. 5:58that is probably surprising to many.
  169. 5:59However, for Entropic, I am making an
  170. 6:02exception, as I said, and will invest a
  171. 6:04very significant portion of my cash
  172. 6:06reserve into the company. Crucially, I
  173. 6:09have subscribed, but only to about 20%
  174. 6:12of the total capital I plan to invest in
  175. 6:15the company.
  176. 6:17And there are two main reasons for this
  177. 6:18. First, and this sounds very grand, for
  178. 6:21me, Entropic is the stock market bet of
  179. 6:24my life.
  180. 6:26And that is primarily because I truly
  181. 6:28view it as a long-term investment.
  182. 6:31I see this investment over 15, 20, 25
  183. 6:33years or even longer, and with such a
  184. 6:35long investment horizon, price
  185. 6:37fluctuations in the first three, four,
  186. 6:40five days, or the first two weeks should
  187. 6:42have no influence on the total
  188. 6:44investment success after 20 years.
  189. 6:47And if they do, then it was the wrong
  190. 6:49investment anyway. And secondly,
  191. 6:51honestly, I don't like this subscription
  192. 6:53hack nonsense at all.
  193. 6:55What is happening right now with the
  194. 6:57brokers, in the media, how it is being
  195. 6:59promoted,
  196. 7:01I don't like it. I'll say it quite
  197. 7:03openly.
  198. 7:04Of course, there is a lot of talk about
  199. 7:06the opportunities, and we do that too,
  200. 7:08but in my opinion, many important things
  201. 7:10are being overlooked.
  202. 7:13Perhaps not even maliciously, but I
  203. 7:14believe everyone who has already
  204. 7:16subscribed or is thinking about
  205. 7:17subscribing should know this.
  206. 7:20Specifically, it is about the price at
  207. 7:22which you can subscribe.
  208. 7:24You surely know that Entropic itself has
  209. 7:27issued a range of 70 to 78 dollars for
  210. 7:29what the share should cost if you
  211. 7:32subscribe.
  212. 7:33So far so good, but the prospectus also
  213. 7:36states that the price can be 20%higher
  214. 7:39without the need for any further
  215. 7:41announcement. So it is possible that if
  216. 7:45you subscribe without a limit, you end
  217. 7:47up paying 93.60 cents for the share.
  218. 7:51That is a significant difference, and
  219. 7:53then there is another possibility, which
  220. 7:55BaFin has also approved, that applies to
  221. 7:58European investors: the price could rise
  222. 8:00to 120 dollars upon subscription.
  223. 8:04For that to happen, there would need to
  224. 8:05be an extra announcement, a supplement,
  225. 8:07but it is possible.
  226. 8:09It is unlikely, but it could happen,
  227. 8:11especially since the stock is naturally
  228. 8:13also heavily oversubscribed.
  229. 8:15That is not our most likely case, but I
  230. 8:18personally assume that the stock will be
  231. 8:21more expensive than 78 dollars, even if
  232. 8:24you subscribe, and I think you should
  233. 8:26just know that.
  234. 8:28So my plan is relatively clear. I will
  235. 8:30take the first small tranche now and try
  236. 8:32to invest it at the start.
  237. 8:35It is also not certain that the order
  238. 8:37will go through in the end, so I will
  239. 8:38invest the rest, the majority of the
  240. 8:40planned capital, in the coming days,
  241. 8:42weeks, and months.
  242. 8:44This is a good point to mention that
  243. 8:46this is just my plan, it is my money,
  244. 8:48and this is not investment advice.
  245. 8:51Speaking of risks, after my part, after
  246. 8:53this section here, Mario will go through
  247. 8:55the EnShop figures with you.
  248. 8:58We really dove deep into the financials
  249. 9:01, some great number crunching with the
  250. 9:03team, where all the classic, let's say,
  251. 9:06stock-specific risks and opportunities
  252. 9:08of EnShop are listed, along with various
  253. 9:11future scenarios we believe are possible
  254. 9:14in the coming years.
  255. 9:17I personally approach this topic from a
  256. 9:19slightly different perspective, which is
  257. 9:21why I am making this exception.
  258. 9:24But before I explain that to you, it is
  259. 9:26important to me to say that with all the
  260. 9:28enthusiasm, there is one thing we must
  261. 9:30never forget.
  262. 9:33In the long run, the biggest influence
  263. 9:35on stock market prices is simply profits
  264. 9:38. You can see it here. Over the long
  265. 9:40term, profits and stock prices always
  266. 9:43align, and EnShop will likely be no
  267. 9:46exception in the long run.
  268. 9:50I want to talk to you now about why I
  269. 9:53believe those very profits could sprout
  270. 9:55to a great extent at EnShop in the
  271. 9:58future, and to begin, I would like to
  272. 10:00quote someone.
  273. 10:03And that is Peter Lynch, who once said,
  274. 10:05use your personal experience.
  275. 10:08So. We must not make the mistake of
  276. 10:10saying, okay, we will skip fundamental
  277. 10:12analysis entirely.
  278. 10:14As I said, Mario will take care of that
  279. 10:16shortly, but I think this approach is
  280. 10:18important—that you don't just look at
  281. 10:20the fundamentals, but, as he says
  282. 10:22himself, go through the world with your
  283. 10:24eyes open.
  284. 10:26That is why my approach, the way I
  285. 10:28currently see and evaluate Entropic, and
  286. 10:31why I want it in my portfolio, is this.
  287. 10:35The answer to that is not found in a
  288. 10:38profit margin or a net loss right now.
  289. 10:44The reason for that is here, in the
  290. 10:46office.
  291. 10:48And that is what I want to talk to you
  292. 10:49about. Let’s divide this into two parts
  293. 10:51.
  294. 10:52First, the question of why I am so
  295. 10:53bullish on AI as a whole, as a future
  296. 10:55technology, and then the perhaps even
  297. 10:57more important question of why I am so
  298. 10:59bullish specifically on Entropic and
  299. 11:01Claude.
  300. 11:02But let's start with AI as a whole.
  301. 11:04Currently, the political debate on how
  302. 11:07to contain AI is heating up more and
  303. 11:09more. The Democrats just won the
  304. 11:11midterms, among other things, precisely
  305. 11:13with this topic.
  306. 11:14But it's not just the Democrats; these
  307. 11:16demands and fears are coming from all
  308. 11:18political circles.
  309. 11:20Here, for example, Steve Bannon is also
  310. 11:22demanding that the pace be slowed down,
  311. 11:23and Bernie Sanders is even calling for a
  312. 11:25moratorium on the expansion of AI data
  313. 11:27centers, right? All of this is being
  314. 11:29discussed right now.
  315. 11:31This is also the big concern in the
  316. 11:33stock market, and I believe you can see
  317. 11:34that in the prices over the last few
  318. 11:36weeks and months. However, there is a
  319. 11:38simple reason why I am extremely relaxed
  320. 11:40about this.
  321. 11:42This whole thing is not new. History
  322. 11:43repeats itself. First example.
  323. 11:47In 1865, the British Parliament passed
  324. 11:49the so-called Red Flag Act to, quote,
  325. 11:52restrict the use of steam-powered and
  326. 11:54other mechanically propelled road
  327. 11:56vehicles, yes?
  328. 11:59These were practically the predecessors
  329. 12:01of the automobile as we know it today,
  330. 12:03with completely wild regulations like:
  331. 12:06within towns, in the city, such a
  332. 12:08vehicle could only travel at a maximum
  333. 12:10of 3.2 km/h, or that three people always
  334. 12:13had to be present to operate the vehicle
  335. 12:15.
  336. 12:17And therefore: Red Flag Act—a man always
  337. 12:20had to walk at least 60 yards in front
  338. 12:23and wave a red flag to warn carriages
  339. 12:25and horses.
  340. 12:27Yes, even back then, it was about
  341. 12:29protecting the horse industry and horse
  342. 12:32breeding.
  343. 12:33Didn't quite work out. The next example
  344. 12:36only a little later.
  345. 12:37In 1888, the former Governor of New York
  346. 12:40State, Alonso B. Cornell, called on the
  347. 12:42Mayor of New York to enforce a ban on
  348. 12:45high-voltage electrical systems in New
  349. 12:47York City.
  350. 12:49A year later, it actually happened, and
  351. 12:51it must be said, after a few deaths, it
  352. 12:53was a dangerous endeavor back then with
  353. 12:55those high-voltage lines.
  354. 12:58The law actually came into effect. Uh,
  355. 13:00in 1889, large parts of the electrical
  356. 13:03street lighting in New York were
  357. 13:05actually switched off.
  358. 13:07Today, they are lit up again.
  359. 13:10And the last example isn't that long ago
  360. 13:11, in 1996 President Bill Clinton signed
  361. 13:13the so-called Telecommunications Act of
  362. 13:151996.
  363. 13:18It included a law that was officially
  364. 13:20intended to protect minors from
  365. 13:21pornographic and offensive content on
  366. 13:23the internet.
  367. 13:25We know the current discussion, um, from
  368. 13:27the EU.
  369. 13:29But the measures were so advanced and so
  370. 13:31comprehensive that even lawful
  371. 13:33communication between adults could have
  372. 13:35been restricted.
  373. 13:37However, I find the reasoning behind why
  374. 13:39it was to be introduced particularly
  375. 13:41interesting. Here is a quote from
  376. 13:43Republican Senator Dan Coats, who said
  377. 13:45at the time,
  378. 13:52Well, sometimes our technology develops
  379. 13:54faster than we can think about its
  380. 13:56consequences, and that is exactly the
  381. 13:57situation we have now with the internet
  382. 13:59.
  383. 14:02Sounds familiar. What do all three
  384. 14:03examples have in common now?
  385. 14:06Despite massive initial headwinds, both
  386. 14:09politically and socially, they
  387. 14:11ultimately prevailed to an incredible
  388. 14:13extent.
  389. 14:16The automobile, as well as electricity,
  390. 14:18as well as the internet.
  391. 14:20And why? Because they are disruptive
  392. 14:23technologies.
  393. 14:25And throughout human history, disruptive
  394. 14:27technologies have always prevailed in
  395. 14:29the end.
  396. 14:31What exactly is a disruptive technology
  397. 14:33? There are six characteristics. First,
  398. 14:35it has a superior economic benefit.
  399. 14:38That means it solves a problem faster,
  400. 14:39easier, better, and cheaper than
  401. 14:41anything that came before it.
  402. 14:43Second, it continuously improves and its
  403. 14:46costs decrease. Third, it is scalable.
  404. 14:48Fourth, it has broad applicability. That
  405. 14:51means it doesn't just apply to one
  406. 14:52product, but to entire economic sectors
  407. 14:54.
  408. 14:56Fifth, it has a creative power of
  409. 14:58destruction, because it destroys entire
  410. 15:01industries, creates new markets, etc.,
  411. 15:03and forces all market participants to
  412. 15:05adapt.
  413. 15:07And sixth, it has a self-reinforcing
  414. 15:09effect.
  415. 15:11So, the more money that flows in, the
  416. 15:13more innovation that flows in, the
  417. 15:14better it gets.
  418. 15:16And I think this is not a hot take that
  419. 15:19I have.
  420. 15:21AI is exactly that. And that is exactly
  421. 15:24why I am bullish on the AI sector in the
  422. 15:25long term.
  423. 15:27However, this does not mean, and this is
  424. 15:29important, that I assume the stock
  425. 15:30prices of all AI stocks—and also
  426. 15:32Entropic—will now only go up in a
  427. 15:34straight line.
  428. 15:36No. There can be massive price drops in
  429. 15:39the future, and they will likely come.
  430. 15:44And there, too, a look at the past is
  431. 15:45worthwhile. We have had this before, too
  432. 15:47. The dot-com bubble. The internet did
  433. 15:50prevail. We all use the internet.
  434. 15:53But how many companies on the stock
  435. 15:55market were torn apart on that bumpy
  436. 15:57road?
  437. 15:58And it wasn't just companies that
  438. 16:00everyone today naturally knows were
  439. 16:02nothing more than castles in the air
  440. 16:03back then. No, it also tore apart the
  441. 16:06companies that dominate the world today
  442. 16:08.
  443. 16:09Let's take a look at Amazon and Nvidia
  444. 16:10for that.
  445. 16:12Both companies were really hit hard
  446. 16:14after the turn of the millennium—for
  447. 16:16Amazon it was right at the turn of the
  448. 16:18millennium, for Nvidia two years later.
  449. 16:21Anyone who was invested back then
  450. 16:24suffered losses of up to 94.4%. Yes, so
  451. 16:28they practically lost almost their
  452. 16:30entire investment.
  453. 16:32That's how deep the stocks plummeted,
  454. 16:34but that is the point; if you zoom out
  455. 16:37to today, those 90%losses are just a
  456. 16:39small dent, and that is exactly what
  457. 16:41leads us to Anthropic.
  458. 16:43Can the stock fall by 90%in the future?
  459. 16:46Yes.
  460. 16:47Do I have a guarantee that this is the
  461. 16:49exact company that will ultimately
  462. 16:51prevail in the AI sector?
  463. 16:53No. And even if it does, I still don't
  464. 16:55know if that will be reflected in the
  465. 16:57shareholder return.
  466. 17:00I am aware of this risk, and anyone who
  467. 17:02invests there should be aware of this
  468. 17:04risk.
  469. 17:06But why is this risk worth it to me in
  470. 17:07this case?
  471. 17:09For the first time in my adult life, I
  472. 17:12have the desire to be involved in a
  473. 17:15company that is not my own because the
  474. 17:18product excites me so much.
  475. 17:21I'll say it quite openly: I want to be
  476. 17:24part of this company and the development
  477. 17:26it might still take, hopefully over the
  478. 17:28long term—at least 10, 15, 20 years,
  479. 17:31perhaps even longer.
  480. 17:33I grew up with the internet, the
  481. 17:34automobile, and electricity. I already
  482. 17:37had all of that in abundance, which is
  483. 17:39why I never had that "wow" effect.
  484. 17:42And in my perception, social media was
  485. 17:44also more of an evolution than a
  486. 17:46disruption.
  487. 17:49For the first time, I am now
  488. 17:51experiencing the incredible power of a
  489. 17:53disruptive technology through Claude,
  490. 17:56the AI from Anthropic, and I can
  491. 17:58practically use it profitably in my
  492. 18:01everyday life and in my company.
  493. 18:05I am not an AI expert. Some of you who
  494. 18:07are much deeper into the topic might
  495. 18:09smile now or write a snarky comment.
  496. 18:13There is so much more than what I am
  497. 18:15currently doing with Claude.
  498. 18:18I am learning, but I would like to use a
  499. 18:21concrete example to show you what is
  500. 18:23already possible in our company so that
  501. 18:26you can understand my fascination.
  502. 18:29Mario and I come, as some of you might
  503. 18:31know, from traditional print journalism
  504. 18:33.
  505. 18:33Today, to finish a magazine story, you
  506. 18:36need significantly more than just an
  507. 18:38editor who provides continuous text.
  508. 18:40You need an art director who plans the
  509. 18:42layout.
  510. 18:43Then, if you can afford it, you need a
  511. 18:45layout artist to implement the plan,
  512. 18:47setting the columns, deciding where the
  513. 18:49images go, and so on.
  514. 18:52You need an infographic designer to
  515. 18:53recreate the charts in your own
  516. 18:54corporate identity.
  517. 18:56You need a photo editor to search
  518. 18:58databases for suitable images, and of
  519. 19:00course, a final editor to proofread all
  520. 19:02the texts and ensure everything is
  521. 19:03structured properly.
  522. 19:05So far, so good.
  523. 19:07When Mario and I founded beating beta
  524. 19:09and became self-employed, it became
  525. 19:11clear quite quickly that we also wanted
  526. 19:13to offer stock reports.
  527. 19:15Not printed, heaven forbid, but rather
  528. 19:18as PDFs.
  529. 19:19Our problem was that while our content
  530. 19:22was very good—and I still stand by that
  531. 19:24today, our stock reports were great from
  532. 19:27the start in terms of content—that was
  533. 19:30all they were, because, layout-wise, we
  534. 19:32neither had the capital nor the team to
  535. 19:35compete with the big editorial offices.
  536. 19:39That was the harsh truth, and the result
  537. 19:41was that I said, "Okay, I'll take care
  538. 19:44of the entire design process."
  539. 19:46I did it all myself, but I had never
  540. 19:48learned how, and those who have been
  541. 19:50with beating beta from the start or for
  542. 19:52a long time know what those reports
  543. 19:54sometimes looked like in terms of layout
  544. 19:56.
  545. 19:59And now, we have Claude. What did we
  546. 20:00actually do? We went to a design agency
  547. 20:03and said, "Please build us master pages
  548. 20:06."
  549. 20:07Various variations that we can use for
  550. 20:09our stock reports, for example,
  551. 20:11two-column, single-column, with charts,
  552. 20:14lead images, stock info boxes, etc., so
  553. 20:16that we have a high-quality base design
  554. 20:19to build upon that is variable enough to
  555. 20:21swap things out.
  556. 20:24Every editorial office works this way,
  557. 20:26every publisher works exactly like this
  558. 20:29: having a master layout with sample
  559. 20:31pages that can be flexibly adapted
  560. 20:33depending on what is needed.
  561. 20:37We also asked for a standardized chart
  562. 20:39style to be developed, so that all
  563. 20:42charts in the Beating Beta design are
  564. 20:44well-structured and clean, and in the
  565. 20:47end, we settled on a design that would
  566. 20:49be the official Beating Beta chart
  567. 20:52design.
  568. 20:54So, we took all of that and gave it to
  569. 20:57Claude, and together with Claude, I
  570. 21:00built a skill that now does the
  571. 21:02following. So, you have to imagine it
  572. 21:05like this: every editor can now simply
  573. 21:08write their copy in a shared Google Doc
  574. 21:11, just hack away, formatting doesn't
  575. 21:14matter at all, just write a clean
  576. 21:17journalistic text into these Google Docs
  577. 21:20and any charts you want—if you find
  578. 21:23something from JP Morgan or wherever—
  579. 21:26just screenshot them and dump them into
  580. 21:29this script, and then the magic begins.
  581. 21:33Because as soon as the body text is
  582. 21:35finished, I briefly notify Claude and
  583. 21:36tell it, "Based on this text, give me a
  584. 21:38layout proposal for our stock report,"
  585. 21:40and Claude does exactly that, slips into
  586. 21:42the role of the art director, and
  587. 21:44presents me with a template.
  588. 21:46I can then flip through the individual
  589. 21:48pages and give it feedback, very
  590. 21:50detailed, I can say I want this there,
  591. 21:51I'd rather have it this way there, I
  592. 21:53like this, I don't like that yet, and
  593. 21:55the adjustments are made in a flash.
  594. 21:58Furthermore, Claude really looks at
  595. 21:59every chart that we simply drop into the
  596. 22:01raw script as a screenshot.
  597. 22:03It interprets this chart, lets us check
  598. 22:06it, and then rebuilds it according to
  599. 22:08the design agency's specifications,
  600. 22:11following our master design in our CI.
  601. 22:13And it looks like this: I’ve brought
  602. 22:15along an original screenshot here, very
  603. 22:17classic, the way we often use them. And
  604. 22:20then I can say, "Uh, Claude, please make
  605. 22:22this a two-thirds layout, also as an
  606. 22:24animation."
  607. 22:25Then I can say, "Hey, for the China
  608. 22:27Report, I'd rather have a red Chinese
  609. 22:29background." Then the whole thing looks
  610. 22:31like this. Or I can also say, I need it
  611. 22:33in 9:16 for social media.
  612. 22:35All in seconds. But that’s still not all
  613. 22:38, because Claude also takes on the role
  614. 22:39of photo editor and final copy editor.
  615. 22:42We have linked Claude to the large image
  616. 22:44databases, and for every spot in the
  617. 22:46report where an image is intended,
  618. 22:48Claude searches for three images for me
  619. 22:51. And then I have the free choice and
  620. 22:53can say, okay, I like image A, B, or C
  621. 22:54the best. I just click on it, the image
  622. 22:56is then automatically inserted and also
  623. 22:58licensed, and if I don't like any of the
  624. 23:00images, it just keeps searching.
  625. 23:02And the final approval is also handled
  626. 23:05by the AI: it checks all texts for
  627. 23:07spelling errors, for anything that
  628. 23:09doesn't flow well, for logical errors;
  629. 23:11it makes sure the breaks look nice and
  630. 23:13the body text is fluid.
  631. 23:16And the whole thing takes—I timed it—
  632. 23:19once the body text, the journalistic
  633. 23:21part is finished and the script is
  634. 23:23handed over to Claude, the entire design
  635. 23:26process takes
  636. 23:31on average just 12 minutes and 41
  637. 23:32seconds.
  638. 23:34Not even 13 minutes. For comparison,
  639. 23:37before Claude and without this skill,
  640. 23:41the same process took us about a week.
  641. 23:45Overnight, we became competitive in
  642. 23:48terms of design and layout with the
  643. 23:50major publishers, and with much less
  644. 23:52staff.
  645. 23:53But even more important to me is that
  646. 23:56this 98%time saving makes our product
  647. 23:59better, because we now have much more
  648. 24:02time for research.
  649. 24:05We have much more time to discuss stocks
  650. 24:07as a team, and above all, we can react
  651. 24:10to the market much faster.
  652. 24:12When you have long design and production
  653. 24:15chains, it’s not exactly ideal if,
  654. 24:18shortly before a planned release, a
  655. 24:20company being featured makes some kind
  656. 24:23of announcement and things change; you
  657. 24:26can barely intervene. Yes, the classic
  658. 24:29horror story is publishers who still
  659. 24:33rely on print.
  660. 24:35Once it's in final editing and at the
  661. 24:36print shop, you can't change anything,
  662. 24:38and sometimes days pass in between.
  663. 24:41So, an article about Stock XY hits the
  664. 24:44newsstand, but it’s no longer current.
  665. 24:48The longer your processes are, the worse
  666. 24:51your ability to react.
  667. 24:54Now, we can swap things out up to two
  668. 24:56minutes before a scheduled publication.
  669. 24:59We can drop a stock, add a new one, we
  670. 25:01can insert or remove paragraphs.
  671. 25:04That gives us complete new freedom and
  672. 25:06makes the product better.
  673. 25:08We have more time to react and more time
  674. 25:10for creative tasks.
  675. 25:13I, for example, can sit here today—
  676. 25:15which might not be an improvement—
  677. 25:17simply because we made those repetitive
  678. 25:20design tasks 98%more efficient.
  679. 25:23And I'm sitting here even though we have
  680. 25:26two reports in final approval tomorrow.
  681. 25:29So, tomorrow we will finalize two
  682. 25:31reports.
  683. 25:32That would have been unthinkable just
  684. 25:35two weeks ago. And that is just one
  685. 25:37example of many.
  686. 25:39About a month ago, I told myself, "
  687. 25:41Sinan, go through every area of the
  688. 25:44company, talk to the person in charge,
  689. 25:47and work with them to automate all
  690. 25:49repetitive tasks—simply put, all the
  691. 25:52tasks nobody wants to do."
  692. 25:56And it’s working wonderfully.
  693. 25:59The team is happy because they suddenly
  694. 26:01have time to focus on the things that
  695. 26:03actually fulfill them.
  696. 26:06Another example is video editing. Yes,
  697. 26:08we produce a lot of videos.
  698. 26:11The rough cut—removing pauses, cutting
  699. 26:13out bad takes—the AI does that now, much
  700. 26:16, much, much faster and error-free.
  701. 26:20And Francesco suddenly has time to focus
  702. 26:22on the enjoyable parts of his job.
  703. 26:26And again, all of this works not thanks
  704. 26:29to me, but in spite of me.
  705. 26:33I’m not even good at prompting, and I
  706. 26:35figure if I can do it, then anyone can.
  707. 26:39And if I just extrapolate the AI
  708. 26:42development of the last few years for
  709. 26:46the next year or two, I don't even want
  710. 26:50to imagine what that means. Because
  711. 26:54today—if I’m creating these skills today
  712. 26:57, I have to dig into it—and I had to
  713. 26:59spend two or three days digging into
  714. 27:02Claude at the beginning.
  715. 27:05I believe in a year or two, no one will
  716. 27:07have to do that anymore.
  717. 27:10In a year or two, it will be so
  718. 27:11intuitive and the barrier so low that
  719. 27:13the processes I’m setting up for our
  720. 27:15company here will be doable for everyone
  721. 27:17.
  722. 27:19And that fascinates me.
  723. 27:21>> Let’s go all in now when it comes to the
  724. 27:23numbers and the valuation. What do we
  725. 27:25like about Anthropic when we look at the
  726. 27:26numbers? We already mentioned this 900%
  727. 27:28before. The customer growth is brutal.
  728. 27:32We see the enterprise customer base here
  729. 27:34. It's insane. Just recently, in
  730. 27:37December 2023, we had 20 here paying
  731. 27:39more than 100k, more than 100,000
  732. 27:42dollars, then 200, 500 in December, more
  733. 27:45than 1,900, and that is obviously the
  734. 27:47big, gigantic lever. Large customers who
  735. 27:50are spending more and more money. At
  736. 27:52first glance, it still looks shaky.
  737. 27:54We’ve already talked about the loss and
  738. 27:55the convertible bonds. At second glance
  739. 27:57, it looks significantly better. Not
  740. 27:59just brutal growth, but Anthropic is
  741. 28:01sitting on cash. We have an operating
  742. 28:03cash flow of plus 3.9 billion dollars in
  743. 28:06the first half of 2026. We have cash
  744. 28:08reserves of more than 131 billion
  745. 28:10dollars before the IPO. So, as you can
  746. 28:13see, everything is relatively solid.
  747. 28:17Now let’s look at it a bit more
  748. 28:18critically, because many fear that this
  749. 28:20is obviously the ultimate hype and in
  750. 28:21two or three years no one will use
  751. 28:22Anthropic anymore, because everyone can
  752. 28:24do AI, and AI might well be successful.
  753. 28:27That is a valid point of criticism, but
  754. 28:28open source and the like—yes, maybe the
  755. 28:30Chinese or someone else will do it all
  756. 28:31soon.
  757. 28:32What exactly is the moat? And there the
  758. 28:35question is, why do customers pay more
  759. 28:36if AI becomes a commodity? Quite simple
  760. 28:38, there are certain lock-in effects, of
  761. 28:40course.
  762. 28:41If I build my company on Claude now,
  763. 28:43then I use these agents, build
  764. 28:44everything around it.
  765. 28:47And now just imagine, yes, there is an
  766. 28:49AI that might be a bit cheaper, even if
  767. 28:51it were a bit better—but it would have
  768. 28:53to be that first—would I really move
  769. 28:54everything over?
  770. 28:56Do I risk everything, the infrastructure
  771. 28:58I’ve built, and then move it all?
  772. 29:00Definitely questionable, to say the
  773. 29:02least.
  774. 29:02We’re seeing lock-in effects, of course
  775. 29:04; we see it in our own company with the
  776. 29:05agents, what’s already possible, and
  777. 29:07naturally, you’re building it all on
  778. 29:08Claude.
  779. 29:09You have a brand. Claude then stands for
  780. 29:11trust. Claude improves itself, which is
  781. 29:13obviously important too. We have
  782. 29:15reliability; customers are paying for
  783. 29:17the work done now, not just for the
  784. 29:18computing power, and of course, we have
  785. 29:20the scale.
  786. 29:21We have a head start, that is crystal
  787. 29:23clear. We are even seeing a trend in the
  788. 29:25US now where they’re hitting the brakes
  789. 29:27, politically as well, and that actually
  790. 29:30plays right into the hands of the
  791. 29:32players who are already big, who have
  792. 29:34effectively made it, and who have the
  793. 29:36computing power. Of course there are
  794. 29:38risks, we need to be very clear about
  795. 29:40that.
  796. 29:41If we look at the business, sure, there
  797. 29:43could be a price drop; it getting
  798. 29:45cheaper is actually likely.
  799. 29:48As for growth, it may have already
  800. 29:49peaked—meaning Anthropic is still
  801. 29:51growing strongly, but just a bit more
  802. 29:53slowly.
  803. 29:54The risks, the commitments for data
  804. 29:56centers, those are financial risks, of
  805. 29:58course. Naturally, customers could also
  806. 30:00throttle their usage, that’s completely
  807. 30:03clear, and headwinds from politics, but
  808. 30:05also competition, are definitely
  809. 30:07something to watch.
  810. 30:08Entering now, if I buy Anthropic, sure,
  811. 30:10the high valuation—a lot is already
  812. 30:11priced in, we’ll look at that in a
  813. 30:13moment.
  814. 30:14Then, of course, a certain amount of
  815. 30:16hype; insiders are allowed to sell early
  816. 30:18, and shareholders have very little say
  817. 30:20. And AI carries other risks entirely,
  818. 30:22of course. You only need to look at what
  819. 30:24Anthropic itself published for its IPO.
  820. 30:26Claude has already acted unintentionally
  821. 30:28and misguidedly, even taking harmful
  822. 30:29actions without any instruction. That is
  823. 30:31certainly possible; those are risks,
  824. 30:33that must be stated very clearly. AI
  825. 30:35today, 100%? No one can do that without
  826. 30:37enough computing power, no top-tier
  827. 30:40models. The commitments for that add up
  828. 30:42to, as I said, over 900 billion dollars
  829. 30:44.
  830. 30:45Then you’re obviously dependent on major
  831. 30:47clients, that is crystal clear.
  832. 30:50Regarding revenue growth, we’ll have to
  833. 30:52see how the stock market digests that,
  834. 30:54because it won't keep rising forever,
  835. 30:56and the founders, as mentioned, hold
  836. 30:5750.1%of the voting rights in the Trust,
  837. 30:59determine the board majority, so Class A
  838. 31:02shareholders have almost no say.
  839. 31:04You naturally need to know that if you
  840. 31:06invest in this stock. How do we value
  841. 31:08this stock now? We have of course done
  842. 31:09our own thinking and aren't just
  843. 31:11parroting what banks or some analysts
  844. 31:13have calculated; we are doing the math
  845. 31:14ourselves. That is our approach to
  846. 31:16valuation. We are looking at the market
  847. 31:17. There is this 40 trillion dollar
  848. 31:19figure out there, representing what
  849. 31:21knowledge workers worldwide generate in
  850. 31:23a year. That is the market. 80%of that
  851. 31:26is addressable. Capabilities: AI can
  852. 31:29handle 70%of knowledge work by 2030,
  853. 31:32that is the second assumption. Customers
  854. 31:34utilize a good half of that. Then the
  855. 31:36share: per dollar spent on AI, customers
  856. 31:38save 8 to 10 dollars in labor.
  857. 31:4035%of model expenditures then go to
  858. 31:43Anthropic. Number 4: Customers pay 400
  859. 31:45dollars per agent user per month. Prices
  860. 31:47per task fall by 35%per year.
  861. 31:50So we have naturally factored in falling
  862. 31:53prices, and the valuation—you can argue
  863. 31:55about this—here is a 25x operating
  864. 31:57profit for the year 2030, discounted to
  865. 31:59today at 12%per year.
  866. 32:02And then we arrive at the following
  867. 32:04scenarios; if we look ahead to 2030, we
  868. 32:07have our base case, our bull case, and
  869. 32:09our bear case, and as you can see, the
  870. 32:12base case leads to a 453 billion dollar
  871. 32:14run rate by the end of 2030.
  872. 32:17That would be our base scenario in the
  873. 32:19middle.
  874. 32:20Bull scenario 680 billion, and the bear
  875. 32:22scenario, if things go wrong, we would
  876. 32:25be at only 184 billion dollars.
  877. 32:28What does this mean concretely for the
  878. 32:29stock? We also have three scenarios.
  879. 32:31There is a price range for the IPO,
  880. 32:34which could theoretically still be
  881. 32:36adjusted upwards. Be very careful with
  882. 32:39that, please look into the details of
  883. 32:41what can happen when subscribing.
  884. 32:43The price range is 70 to 78 dollars, but
  885. 32:46as I said, that could theoretically
  886. 32:48shift upwards.
  887. 32:50Pay attention to the details there.
  888. 32:51Where do we see the stock now?
  889. 32:53Here you can see the wide range, and
  890. 32:55that is naturally also the massive risk
  891. 32:57.
  892. 32:58In the bear case, we see -74 percent; if
  893. 33:01we assume a median issuance price of 74
  894. 33:04dollars, then we would be at 19 dollars
  895. 33:07in the bear case, 108 dollars in the
  896. 33:10base case, and 247 dollars in the bull
  897. 33:13case. Important: this entire video is of
  898. 33:15course not investment advice, it serves
  899. 33:17information purposes only.
  900. 33:18Please note the massive risks associated
  901. 33:21with this stock and this company;
  902. 33:23neither I nor Sinan Krieger nor the
  903. 33:25channel are liable for potential losses
  904. 33:27.
  905. 33:28So folks, we have now reached the end of
  906. 33:29this video. I hope you liked it and were
  907. 33:31able to take something away from it.
  908. 33:34Feel free to leave a thumbs up,
  909. 33:36definitely subscribe to my channel, and
  910. 33:38the Beating Beta channel as well for
  911. 33:40more analyses, and the interviews with
  912. 33:42the best financial experts in Germany
  913. 33:44are definitely worth it.
  914. 33:46So, I subscribed to the stock as well.
  915. 33:49There are brutal risks, it's quite a bit
  916. 33:51of hype, and there are high expectations
  917. 33:53, but I see what’s possible every day in
  918. 33:55our companies—things that used to be
  919. 33:57really complex, like websites, coding,
  920. 33:59and process automation—where so much is
  921. 34:01happening in the background, requiring
  922. 34:03so many database providers, which can
  923. 34:05all more or less be taken over by cloud
  924. 34:07soon.
  925. 34:09It’s absolutely insane, and we in the
  926. 34:12company are naturally prepared to spend
  927. 34:14more money on it because it simply saves
  928. 34:17a massive amount of money in other areas
  929. 34:19; I mean, we aren't firing any employees
  930. 34:22.
  931. 34:23Our employees are very important to us,
  932. 34:25but where we might have needed new hires
  933. 34:27before, I don't need them now, and I’d
  934. 34:29rather spend that money on AI because
  935. 34:31the productivity and efficiency gains
  936. 34:33are just brutal; it’s really wild.
  937. 34:37And that’s why I see Cloud very well
  938. 34:39positioned and definitely think Cloud is
  939. 34:41better than other AIs; when you’re
  940. 34:43familiar with it from daily use, it’s
  941. 34:45really something else.
  942. 34:47But I also think, well, I was also very
  943. 34:49cool-headed about the SpaceX IPO.
  944. 34:52I definitely believe that you will be
  945. 34:53able to get this stock for a better
  946. 34:55price later.
  947. 34:57We are already in a somewhat difficult
  948. 34:58environment for AI, so please don't be
  949. 35:00greedy, please no FOMO, please proceed
  950. 35:02with a plan, and if you prefer to stay
  951. 35:04away, that is of course completely fine
  952. 35:05.
  953. 35:06You shouldn't feel too much pressure
  954. 35:08about it either.
  955. 35:09So, thank you all for watching.
  956. 35:13It was a great honor, and I hope you
  957. 35:15enjoyed it.
  958. 35:16You can find our advent calendar link
  959. 35:18below, where you can win great prizes,
  960. 35:20and if you want to see our full Entropic
  961. 35:21analysis for our members, you can still
  962. 35:23sign up until the 30th.
  963. 35:25November to test our Daily Briefing for
  964. 35:28free, just click up here.
  965. 35:32And here you can watch my last briefing
  966. 35:37again.

About this transcript

This page contains the full transcript of Anthropic-IPO: Ich zeichne! // die große Analyse der Claude-Aktie by Mario Lochner, generated from the public captions YouTube serves with the video. The transcript has 6,013 words across 966 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.