YouTube2Text

Colette Kress takes charge of Nvidia: how its AI bet could change — Transcript

by Finance for everyone · 3,683 words · 606 segments · language en · Watch on YouTube

Full transcript

  1. 0:00This is Colette Kress, and she just took
  2. 0:02charge of NVIDIA, the company that makes
  3. 0:04the chips behind most of the world's
  4. 0:06artificial intelligence. And she doesn't
  5. 0:08come from designing chips.
  6. 0:11Until now, she had a different job, and
  7. 0:14NVIDIA's announcement puts it on the
  8. 0:15same line as her promotion.
  9. 0:17Colette Kress, Executive Vice President
  10. 0:20and Chief Financial Officer, appointed
  11. 0:22interim CEO.
  12. 0:24The Chief Financial Officer is the one
  13. 0:26who runs the books. She controls how
  14. 0:28much gets invested, what every dollar is
  15. 0:30spent on, and how much the company makes
  16. 0:33on each sale.
  17. 0:35And interim [music]
  18. 0:36is because Jensen Huang, the founder who
  19. 0:38ran NVIDIA for 33 years, left overnight,
  20. 0:42and the board is still looking for
  21. 0:43someone permanent. In the meantime,
  22. 0:45she's doing both jobs at once.
  23. 0:48NVIDIA used to be run by the engineer
  24. 0:50who founded it, and now it's run by the
  25. 0:53person who knows its numbers best.
  26. 0:55And that raises a huge question. Will
  27. 0:57she keep her foot on the gas with the
  28. 0:59same bet, or will she change what NVIDIA
  29. 1:02sells, who it sells to, and on what
  30. 1:04terms?
  31. 1:06That question breaks down into three.
  32. 1:08NVIDIA has already announced its next
  33. 1:10chips, and its customers are counting on
  34. 1:12them. Will it keep launching them at the
  35. 1:14same pace, or will a CFO prefer to slow
  36. 1:17down and deliver without mistakes?
  37. 1:20Right now, NVIDIA isn't even counting on
  38. 1:22selling its AI chips in China because
  39. 1:25that depends on what Washington allows.
  40. 1:27Can Kress open that door?
  41. 1:30And her job is interim because the board
  42. 1:32is looking for someone else.
  43. 1:34How much can someone who's just passing
  44. 1:35through really change?
  45. 1:38Those are the questions we'll answer
  46. 1:39throughout this video, and their answers
  47. 1:41shape what's at stake for NVIDIA's
  48. 1:43stock.
  49. 1:45And this affects you more than you might
  50. 1:46think. If you own a fund that tracks the
  51. 1:49US stock market, part of your money is
  52. 1:51in NVIDIA.
  53. 1:53But it's not just shareholders because
  54. 1:55its chips shape how much the AI you use
  55. 1:57every day costs and how fast it to you.
  56. 2:01The first clue comes on Tuesday,
  57. 2:03November 17th, when she reports earnings
  58. 2:05for the first time as the boss.
  59. 2:09To understand how she thinks, you first
  60. 2:11need to know what Nvidia sells.
  61. 2:14A regular computer does calculations one
  62. 2:16after another very fast. [music]
  63. 2:18Nvidia's chips do thousands at the same
  64. 2:20time, and that's exactly what an AI
  65. 2:23needs to learn and to answer you.
  66. 2:25Nvidia calls this accelerated computing.
  67. 2:30In 2024, back when she was only running
  68. 2:32the finances, she explained in an
  69. 2:34interview how she sees the future of
  70. 2:36this business.
  71. 2:38We are not going to get this all done in
  72. 2:39two years. These transitions are two
  73. 2:42decades, three decades of work to change
  74. 2:45to accelerating computing and to infuse
  75. 2:47AI in everything that we do.
  76. 2:50Do we know what everyday is going to be?
  77. 2:52Is it all going to grow at this same
  78. 2:54lovely speed? No. But we know it is
  79. 2:57going to be with us for decades.
  80. 3:00In other words, she's not promising that
  81. 3:02sales will keep skyrocketing every year.
  82. 3:05She's promising that this business is in
  83. 3:07it for the long haul with good years and
  84. 3:09weaker [music] years.
  85. 3:11And that comes from the person who now
  86. 3:13decides how much Nvidia invests and in
  87. 3:16what.
  88. 3:17In that same interview, she talked about
  89. 3:19who pays for all this. The companies
  90. 3:22that buy AI.
  91. 3:24Her answer has two parts. The first,
  92. 3:26making the investment now, knowing that
  93. 3:28it's not an instantaneous solution.
  94. 3:31And the second says where the money
  95. 3:33comes from.
  96. 3:34You will stop doing things that are not
  97. 3:36providing productivity.
  98. 3:38Picture an insurance company that spends
  99. 3:40a million dollars a year on people who
  100. 3:42review accident claims by hand.
  101. 3:44It decides to bring in an AI which costs
  102. 3:47$200,000 a year.
  103. 3:50In the first year, it doesn't save
  104. 3:51anything yet because people have to
  105. 3:53learn how to use it. So that year, it
  106. 3:55spends more than before.
  107. 3:57But in the second year, the AI handles
  108. 4:00half the reviews and the insurer only
  109. 4:02needs $500,000 for the ones still done
  110. 4:04by hand. Add the 200,000 for the AI, and
  111. 4:08it spends 700,000 in total, less than
  112. 4:11the million it started with.
  113. 4:13Its budget didn't go up. It moved money
  114. 4:16from manual reviews to the AI. And if
  115. 4:18the math hadn't worked out, it wouldn't
  116. 4:20pay for it again.
  117. 4:22And Nvidia depends on that math. Today,
  118. 4:25its biggest buyers are the big tech
  119. 4:27companies, which fill entire warehouses
  120. 4:30with its chips. But the way Kress sees
  121. 4:32it, the business only lasts for decades
  122. 4:34if thousands of ordinary companies, like
  123. 4:36that insurer, find that AI pays off for
  124. 4:39them.
  125. 4:40Will she look there for Nvidia's next
  126. 4:42wave of growth?
  127. 4:44As the boss, her first words came in the
  128. 4:47same announcement that named her.
  129. 4:49We will continue advancing the work that
  130. 4:51has made Nvidia a leader in accelerated
  131. 4:53computing and AI and execute on the
  132. 4:56opportunities ahead. Our focus remains
  133. 4:58on serving our customers, partners,
  134. 5:01developers, and shareholders.
  135. 5:03It's what anyone stepping in on an
  136. 5:05interim basis would say. Don't panic.
  137. 5:08Everything carries on.
  138. 5:10She doesn't announce a single change.
  139. 5:12But she doesn't say what those
  140. 5:13opportunities are or how she plans to
  141. 5:16seize them either.
  142. 5:18And the first test of that continuity is
  143. 5:20the chips Nvidia has already promised
  144. 5:22its customers. Will it keep launching a
  145. 5:24new one at full speed or will she focus
  146. 5:27on delivering on time what it's already
  147. 5:29sold?
  148. 5:31What Nvidia sells today isn't individual
  149. 5:33chips. It's machines like this one
  150. 5:35packed with its Blackwell chips, which
  151. 5:37big tech companies install in rows in
  152. 5:39their data centers, warehouses full of
  153. 5:42computers where AI runs.
  154. 5:45Until 2024, Nvidia released a new
  155. 5:48generation of chips every two years.
  156. 5:51That year, Jensen Huang announced that
  157. 5:53from then on it would release one every
  158. 5:55year.
  159. 5:56For its customers, it's a race. Each
  160. 5:59generation does more work than the last,
  161. 6:01and whoever doesn't keep up falls behind
  162. 6:03the competition. And for Nvidia, every
  163. 6:06new generation is another round of
  164. 6:08orders.
  165. 6:10But moving that fast comes at a price.
  166. 6:12In 2024, Blackwell came out with a
  167. 6:15design flaw. The chips worked, but each
  168. 6:18batch that was made produced very few
  169. 6:20good ones. Nvidia had to redo part of
  170. 6:22it, and deliveries to customers like
  171. 6:24Meta, Google, and Microsoft were delayed
  172. 6:27by months.
  173. 6:28Huang himself admitted it. It was 100%
  174. 6:32Nvidia's fault.
  175. 6:34Now it's the next generation's turn.
  176. 6:36Vera Rubin, which has already started
  177. 6:38being assembled and reaching its first
  178. 6:40customers. Amazon, for example, is
  179. 6:42already counting on it for its data
  180. 6:44centers. This is what Crest inherits,
  181. 6:47chips that are already in motion and
  182. 6:49that customers expect to receive on
  183. 6:51time.
  184. 6:52And in that 2024 interview, she had
  185. 6:54already explained how she allocates
  186. 6:56Nvidia's money,
  187. 6:58planning our capital to design more
  188. 7:00products and build more products with
  189. 7:02one condition. We do not waste a day. We
  190. 7:05don't waste a dime.
  191. 7:08She wants both to move fast and not
  192. 7:11waste money.
  193. 7:13And a delay like Blackwell's means
  194. 7:15losing both at once, days and dollars.
  195. 7:18Nothing she has said points to slowing
  196. 7:20down the pace of a new chip every year.
  197. 7:23What it does point to is watching
  198. 7:25closely that every chip arrives when it
  199. 7:27was promised.
  200. 7:28And for anyone building AI, that matters
  201. 7:31a lot because a faster chip that shows
  202. 7:33up late leaves their data center half
  203. 7:35finished and their new AI models
  204. 7:37waiting.
  205. 7:38Also, Crest has less say here than it
  206. 7:41might seem. The chips are designed by
  207. 7:43the same engineering team that worked
  208. 7:45with Huang, and the board has already
  209. 7:46said the product roadmap will continue
  210. 7:48without interruption.
  211. 7:51But chips are only part of what a
  212. 7:52company needs to use AI.
  213. 7:55Go back to that insurer. It buys the
  214. 7:57chips, and on their own, they do
  215. 7:58nothing. It needs the software that puts
  216. 8:01them to work, someone to update it when
  217. 8:03a new version comes out, and someone to
  218. 8:05secure it so nobody steals its
  219. 8:07customers' data.
  220. 8:09If it doesn't have engineers for that,
  221. 8:11it has to pay somebody.
  222. 8:13Kress said in that interview that most
  223. 8:15companies don't write their own software
  224. 8:17and that the somebody who takes care
  225. 8:19[music] of it is Nvidia.
  226. 8:20We have to keep it secure. We have to
  227. 8:23keep it continually upgraded.
  228. 8:25And Nvidia already has that storefront
  229. 8:28up and running. On its website, it
  230. 8:30offers more than 100 AI models ready to
  231. 8:32run on its chips, and many of them can
  232. 8:34be tried for free.
  233. 8:36But when a company uses them for real,
  234. 8:39it starts paying.
  235. 8:40On Nvidia's price list, a one-year
  236. 8:42subscription to its software with
  237. 8:44support included costs [music] $4,500
  238. 8:48per chip.
  239. 8:49If the insurer has 1,000 chips, that's
  240. 8:52$4.5 million a year, and it pays it
  241. 8:55again every year for as long as it uses
  242. 8:57them.
  243. 8:59And that's the difference a CFO cares
  244. 9:01about most. A chip is sold once. If the
  245. 9:04customer doesn't buy again the next
  246. 9:06year, that money doesn't come back.
  247. 9:09The subscription, on the other hand,
  248. 9:11comes in every year.
  249. 9:13If Kress put more weight on software,
  250. 9:15Nvidia would have steadier revenue even
  251. 9:17in a year when it sold fewer chips.
  252. 9:20She hasn't announced anything like this
  253. 9:21so far, but it's what she argued for
  254. 9:24when she was only running the finances.
  255. 9:27For companies, it cuts both ways. Paying
  256. 9:30Nvidia for the software saves them from
  257. 9:32building their own team of engineers,
  258. 9:34and for many, it's the fastest way to
  259. 9:36get started with AI.
  260. 9:38But the longer they use it, the more
  261. 9:40they depend on Nvidia.
  262. 9:42If one day they want to switch to
  263. 9:43another maker's chips, they have to
  264. 9:45rebuild everything they built on its
  265. 9:47software.
  266. 9:48But before it can charge for software,
  267. 9:50it has to sell the chips. And some of
  268. 9:53Nvidia's biggest buyers can't build
  269. 9:55their data centers on their own. Nvidia
  270. 9:57gives them a hand. [music]
  271. 9:59Will it keep doing that under Kress, or
  272. 10:01will a CFO think twice?
  273. 10:04An AI data center costs so much that
  274. 10:07even Nvidia's biggest customers look for
  275. 10:09someone to back them.
  276. 10:10One of them is OpenAI, the company
  277. 10:13behind ChatGPT.
  278. 10:15To build its facilities, it has to lease
  279. 10:17the land, secure the electricity, and
  280. 10:19put up the building. And whoever leases
  281. 10:21all that to it wants to be sure they'll
  282. 10:23get paid.
  283. 10:25And that's where Nvidia comes in. Its
  284. 10:27summer financial report puts it like
  285. 10:29this.
  286. 10:30In August 2026, we entered into
  287. 10:32guarantees capped at a total of $105
  288. 10:36billion.
  289. 10:37And it spells out when it would have to
  290. 10:39pay.
  291. 10:40Our payment obligations under the
  292. 10:41guarantees are triggered upon certain
  293. 10:43tenant defaults.
  294. 10:46It's like parents co-signing their kids'
  295. 10:48lease. If the kid pays every month, the
  296. 10:50parents don't put in a cent. If the kid
  297. 10:52stops paying, the landlord goes after
  298. 10:55them.
  299. 10:56Here, the kid is OpenAI. The landlord is
  300. 10:59the company building its data center,
  301. 11:01and the parents are Nvidia.
  302. 11:03So what does Nvidia get for taking on
  303. 11:06that risk?
  304. 11:07In return, that data center will be
  305. 11:09filled almost exclusively with Nvidia
  306. 11:11equipment. In other words, it co-signs
  307. 11:14the lease to make sure those facilities
  308. 11:16fill up with its chips.
  309. 11:18If OpenAI pays, it's all sales. If one
  310. 11:21day it can't, Nvidia has to put up
  311. 11:23money, though only part of the rent, not
  312. 11:26the whole construction.
  313. 11:28Inside Nvidia, not everyone saw these
  314. 11:30deals the same way. According to
  315. 11:32Reuters, citing a person familiar with
  316. 11:34the matter, Cress and her staff clashed
  317. 11:37with Huang and the company's sales team
  318. 11:39over the massive and risky financing
  319. 11:41deals.
  320. 11:42In other words, the person now in charge
  321. 11:44was the one pushing back on these deals.
  322. 11:47She hasn't announced any changes so far,
  323. 11:49but she has two options.
  324. 11:52If she toughens them, for example, by
  325. 11:54guaranteeing less or asking for more in
  326. 11:56return, Nvidia risks less money if a
  327. 11:59customer fails.
  328. 12:01But some customers would have a harder
  329. 12:03time building their data centers. Those
  330. 12:05AI projects would move more slowly, and
  331. 12:07so would their chip orders.
  332. 12:10If she keeps them, the facilities go up
  333. 12:12sooner and fill up with Nvidia chips,
  334. 12:14but the risk stays on its books.
  335. 12:18There's another place where Nvidia could
  336. 12:19sell a huge number of chips, China.
  337. 12:23But there, Nvidia doesn't decide,
  338. 12:25Washington does.
  339. 12:27The US government worries that the most
  340. 12:28powerful chips could help China
  341. 12:30militarily, so it decides which ones can
  342. 12:33be sold there and to whom.
  343. 12:36In January, the Commerce Department
  344. 12:38opened the door a crack for one of
  345. 12:40Nvidia's chips. It will review each
  346. 12:42license case by case as long as certain
  347. 12:45security requirements are met.
  348. 12:48In other words, every sale needs a
  349. 12:50permit and the Chinese buyer has to
  350. 12:52prove who it's passing the chips onto.
  351. 12:55Even so, in August, Kress herself made
  352. 12:58it clear how much Nvidia trusts that
  353. 13:00door.
  354. 13:01There is no China data center compute
  355. 13:04revenue in our forward outlook.
  356. 13:06In other words, in the forecasts Nvidia
  357. 13:08makes for the coming months, whatever it
  358. 13:10sells in China in AI chips counts as
  359. 13:13zero.
  360. 13:14And that has a curious upside for
  361. 13:16shareholders. Since Nvidia is no longer
  362. 13:19counting on China, if the door stays
  363. 13:21shut, it loses nothing it was expecting.
  364. 13:24But if Kress manages to open it,
  365. 13:26everything it sells there is a bonus
  366. 13:28that isn't in its forecast today.
  367. 13:31The problem is the long term. While
  368. 13:33Nvidia isn't selling, Chinese customers
  369. 13:36buy chips from homegrown makers like
  370. 13:38Huawei, and every year that goes by,
  371. 13:40those makers get better and win more
  372. 13:42customers.
  373. 13:44So does Kress have anyone to negotiate
  374. 13:47with? Until now, the one who went to the
  375. 13:49White House to make Nvidia's case was
  376. 13:51Huang, and she has no known political
  377. 13:54connections of her own.
  378. 13:56What she does inherit is a tool. In
  379. 13:58August, Nvidia registered the Nvidia
  380. 14:01Corporation Employees Federal Political
  381. 14:03Action Committee in Washington, and the
  382. 14:05official record reads connected
  383. 14:07organization, Nvidia Corporation.
  384. 14:11A committee like this pools money from a
  385. 14:13company's employees to donate to
  386. 14:15politicians campaigns. Lots of big
  387. 14:17companies have one so that Washington
  388. 14:19listens when they need something.
  389. 14:22But it belongs to Nvidia, not to her.
  390. 14:24That kind of relationship with
  391. 14:26politicians is something Kress will have
  392. 14:28to earn on her own.
  393. 14:30And while China stays half open, other
  394. 14:32countries are knocking on Nvidia's door
  395. 14:34because they want their own AI.
  396. 14:37Could that be where it finds the growth
  397. 14:38China won't give it?
  398. 14:40Today, when a company uses AI, it
  399. 14:43usually rents computers over the
  400. 14:45internet from Amazon, Microsoft, or
  401. 14:47Google in facilities that may be in
  402. 14:49another country. That's what we call the
  403. 14:52cloud.
  404. 14:53And more and more governments don't want
  405. 14:55their AI and their citizens' data to
  406. 14:58depend on a handful of American
  407. 14:59companies. They want data centers on
  408. 15:01their own soil, run by companies from
  409. 15:04their own country.
  410. 15:06Kress laid it out in August when she
  411. 15:08presented earnings.
  412. 15:10A country or region can allocate land
  413. 15:12and power directly to a regional cloud
  414. 15:14partner in ways it never would to a
  415. 15:16foreign hyperscaler.
  416. 15:19Hyperscaler is what the industry calls
  417. 15:21giants like Amazon or Microsoft. What
  418. 15:24she's saying is that a government will
  419. 15:26give a local company the land and
  420. 15:28electricity a data center needs,
  421. 15:30something it wouldn't give an American
  422. 15:32one.
  423. 15:33And who does that local company buy its
  424. 15:35chips from? Nvidia.
  425. 15:38So every country that joins in is a new
  426. 15:40buyer, and Nvidia depends less on the
  427. 15:42big tech companies that are its biggest
  428. 15:44buyers today.
  429. 15:46South Korea is one of those countries.
  430. 15:49There, Nvidia has teamed up with the
  431. 15:51country's big phone carrier, SK Telecom,
  432. 15:54and the Korean company Rebellions as
  433. 15:56part of Korea's plan to have its own AI.
  434. 16:00Will this be Kress's big bet? She hasn't
  435. 16:03said anything new since taking over, but
  436. 16:05it's exactly the path she herself laid
  437. 16:07out in August.
  438. 16:09And it changes who gets to build AI, not
  439. 16:12just a few American companies anymore,
  440. 16:14but also a company in India or Malaysia
  441. 16:17backed by its government.
  442. 16:20But there's a limit, and Nvidia admits
  443. 16:22it in its own filings.
  444. 16:24Finally, our business depends on our
  445. 16:26ability to receive consistent and
  446. 16:28reliable supply from our overseas
  447. 16:30partners, especially in Taiwan and South
  448. 16:33Korea.
  449. 16:35Nvidia designs the chips, but it doesn't
  450. 16:37make them. They're made in factories in
  451. 16:39Taiwan, and key parts like memory come
  452. 16:41from South Korea.
  453. 16:43So no matter how many countries want
  454. 16:45chips, Nvidia can only sell as many as
  455. 16:48those factories can make. And if
  456. 16:50something cuts off that supply, like a
  457. 16:52conflict in the region, sales slow down
  458. 16:54for everyone, countries and big tech
  459. 16:57alike.
  460. 16:58With so many fronts open, we're back to
  461. 17:00the question from the start.
  462. 17:02How much can someone who's just passing
  463. 17:04through really change?
  464. 17:07Being interim doesn't mean she can't
  465. 17:09call the shots.
  466. 17:11Nvidia's bylaws, the internal rules that
  467. 17:13say who's in charge of what, allow her
  468. 17:16to hold two jobs at once to begin with.
  469. 17:19Any one person may hold any number of
  470. 17:21offices.
  471. 17:22And on the CEO's powers, the text ends
  472. 17:24like this.
  473. 17:26Shall also perform such other duties and
  474. 17:28have such other powers as the board of
  475. 17:30directors shall designate from time to
  476. 17:32time.
  477. 17:34In other words, how much Cress can
  478. 17:35decide is up to the board, the group
  479. 17:37that represents the shareholders, which
  480. 17:39just named a chair for the first time in
  481. 17:41its history.
  482. 17:43And Nvidia hasn't disclosed what
  483. 17:45specific powers it has given her.
  484. 17:48On top of that, there's a rule that
  485. 17:49weighs on every deal she inherits.
  486. 17:52Unless authorized or ratified by the
  487. 17:54board of directors or within the agency
  488. 17:57power of an officer, no officer, agent,
  489. 18:00or employee shall have any power or
  490. 18:02authority to bind the corporation by any
  491. 18:04contract or engagement.
  492. 18:07So Nvidia's big commitments, like the
  493. 18:09guarantee for OpenAI, are signed with
  494. 18:12the board's approval, and Cress needs it
  495. 18:14to change them too.
  496. 18:16And a signed contract doesn't get torn
  497. 18:18up just because the boss changes. The
  498. 18:20other side has to agree as well.
  499. 18:23Plus, the board is searching for her
  500. 18:24replacement inside or outside the
  501. 18:27company. If it's not her, any big change
  502. 18:30she makes may last only as long as she
  503. 18:32stays in the job.
  504. 18:34That's why in these months the easiest
  505. 18:36thing for her is to adjust what she
  506. 18:37knows best, the money and the terms of
  507. 18:40new deals rather than the whole
  508. 18:41direction.
  509. 18:43But even just by adjusting the money,
  510. 18:45she can change a lot about what Nvidia
  511. 18:47earns, which direction suits it best,
  512. 18:50and what's at stake for its stock with
  513. 18:52each one.
  514. 18:54To know what's at stake for the stock,
  515. 18:55you first need to understand what you're
  516. 18:57paying for when you buy a share.
  517. 19:00Imagine a bakery that makes $10,000 a
  518. 19:02year, and every year it makes 30% more.
  519. 19:06Someone is willing to pay you $300,000
  520. 19:09for it because in a few years it'll be
  521. 19:11making triple.
  522. 19:12But one day it becomes clear that from
  523. 19:14now on it will only grow 5% a year. It
  524. 19:17still makes $10,000, but nobody will
  525. 19:20give you 300,000 anymore. At most, what
  526. 19:23people pay for an ordinary bakery, say
  527. 19:25150,000.
  528. 19:28It hasn't lost a single customer, and
  529. 19:30yet it's worth half.
  530. 19:33It's the same with Nvidia. Its stock
  531. 19:35price already assumes it will keep
  532. 19:37growing very fast. So what moves its
  533. 19:39stock isn't just how much it earns, but
  534. 19:42whether the path Crest chooses makes
  535. 19:44people believe it'll keep growing like
  536. 19:45that.
  537. 19:47The first path is to stay the course and
  538. 19:49deliver. Get Vera Rubin out on time and
  539. 19:52deliver what was promised. That's what
  540. 19:54the market already expects, so there are
  541. 19:56no surprises, good or bad.
  542. 19:59But she herself said the business won't
  543. 20:01always grow at this speed, and the day
  544. 20:03that shows the stock will behave like
  545. 20:05the bakery.
  546. 20:07The second is to charge more for
  547. 20:09software and sell to the countries that
  548. 20:11want their own AI. If it works, Nvidia
  549. 20:14earns steadier money and depends less on
  550. 20:16a few tech giants, and that makes people
  551. 20:18trust that the growth will last.
  552. 20:20The risk is that it takes years to show
  553. 20:23up, and the market doesn't always have
  554. 20:24that kind of patience.
  555. 20:27The third is to get tougher with big
  556. 20:29customers and guarantee less. And Nvidia
  557. 20:32already admits in its filings what could
  558. 20:33happen. Customers may postpone
  559. 20:36purchasing new architectures, and one of
  560. 20:38the reasons it gives is constraints on
  561. 20:40capital to have sufficient funding to
  562. 20:42purchase our products.
  563. 20:44In other words, if a customer is short
  564. 20:46on money, it doesn't buy the new
  565. 20:48generation of chips. If Kress stops
  566. 20:50giving them a hand, Nvidia sells less in
  567. 20:53the short term.
  568. 20:54But if the AI frenzy cools off and a
  569. 20:56customer can't pay, Nvidia will have
  570. 20:59risked much less. It earns less today in
  571. 21:01exchange for fewer scares tomorrow.
  572. 21:05And hanging over all three paths is
  573. 21:07confidence. One analyst summed it up to
  574. 21:09Reuters like this. Putting Kress in the
  575. 21:12interim role makes sense to keep the
  576. 21:13markets calm, but I don't think she is
  577. 21:15the long-term visionary that Nvidia
  578. 21:17needs to continue its current
  579. 21:19trajectory.
  580. 21:20That's an opinion, not a fact, but it
  581. 21:23explains what's at stake for the stock.
  582. 21:25As long as Kress is interim, the market
  583. 21:27will also be watching who comes next.
  584. 21:31The first test already has a date.
  585. 21:33Nvidia will host its earnings call on
  586. 21:35Tuesday, November 17th, and ahead of it,
  587. 21:38it will provide written commentary on
  588. 21:40its third quarter results from Colette
  589. 21:42Kress.
  590. 21:44That's where we'll see which way she
  591. 21:45leans.
  592. 21:47If she confirms Vera Rubin is on
  593. 21:49schedule and little else, she's taking
  594. 21:51the first path. If she talks more about
  595. 21:53software and countries, she's leaning
  596. 21:55toward the second. And if she changes
  597. 21:57the terms of the guarantees for
  598. 21:59customers like OpenAI, she's taking the
  599. 22:01third.
  600. 22:03Nvidia has a new boss, but so far, not a
  601. 22:06new direction.
  602. 22:07That direction won't come from a press
  603. 22:09release, but from the numbers, what it
  604. 22:11spends on, who it sells to, and how much
  605. 22:14risk is taken by the person who knows
  606. 22:16its numbers best.

About this transcript

This page contains the full transcript of Colette Kress takes charge of Nvidia: how its AI bet could change by Finance for everyone, generated from the public captions YouTube serves with the video. The transcript has 3,683 words across 606 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.