Nvidia’s Troubles Could Become America’s Crisis — Transcript
Full transcript
- 0:00I think most people are reading Nvidia’s terrible
- 0:02month backwards. They look at the falling charts
- 0:05and see a story about one company, one
- 0:07missing chief executive, and a luxury problem for
- 0:10rich tech investors. But in reality, if the
- 0:13artificial intelligence boom breaks the way the last
- 0:16tech boom did, the biggest holdings in your
- 0:18retirement fund do not just take a temporary
- 0:21dip, they fall for years. After two thousand,
- 0:24the Nasdaq lost more than three quarters of
- 0:26its value and took fifteen years to get
- 0:29back to even, and in two thousand and
- 0:31eight, the average workplace retirement account that people
- 0:33held all the way through lost about a
- 0:36third of its value in a single year.
- 0:38So with Nvidia down about thirty percent and
- 0:41the man who ran it gone, does the
- 0:43artificial intelligence boom break right here, and how
- 0:46much of your four-oh-one-k goes with it? On
- 0:48November twelfth, twenty twenty-six, Nvidia announced that its
- 0:52chief executive had resigned, an exit that actually
- 0:54took effect the day before, and that its
- 0:57finance chief, Colette Kress, would step in to
- 0:59run the company while the board looks for
- 1:01a permanent replacement. Four days before that announcement,
- 1:04the company had published an email in which
- 1:06he admitted an affair and backed an independent
- 1:09board inquiry; that investigation is not finished, and
- 1:12nothing has been proven. The official announcement gave
- 1:15no reason at all. But investors did not
- 1:17wait for reasons. By the close of trading
- 1:20on November twelfth, Nvidia’s shares stood about thirty
- 1:23percent below their peak of just five weeks
- 1:25earlier. Let us invent two of the people
- 1:27that kind of drop actually lands on, because
- 1:30otherwise this whole thing sounds like abstract digital
- 1:33weather. Meet Carol and her son Ben. Carol
- 1:36is sixty years old, works an ordinary job,
- 1:38and has never bought an individual stock in
- 1:40her life. Her four-oh-one-k holds about a hundred
- 1:43and seven thousand dollars, which is about the
- 1:46median balance at Vanguard for savers aged fifty-five
- 1:49to sixty-four. Her money sits entirely in a
- 1:52single target-date fund designed for someone retiring around
- 1:55twenty thirty, meaning she has five years left
- 1:58before she plans to stop working. Her son
- 2:00Ben is thirty, has about nineteen thousand dollars
- 2:03in the target-date fund for people his age,
- 2:05and has thirty-five years of work ahead of
- 2:07him. Neither of them ever picked a chip
- 2:09stock, but their retirement accounts did it for
- 2:12them. My own impression of that five-week slide
- 2:15is that the market is finally waking up
- 2:17to the real danger, which was never about
- 2:19an executive's personal life. To see why one
- 2:22chipmaker’s bad month reaches accounts like Carol’s, you
- 2:25have to look at where Nvidia’s revenue actually
- 2:27comes from. Most people assume a computer hardware
- 2:30company operates like an appliance store. You build
- 2:33a machine, a customer walks in, pays cash
- 2:36for it, carries it away, and whatever happens
- 2:38to that customer tomorrow is entirely their own
- 2:41business. That is what the word customer usually
- 2:43promises. But in the artificial intelligence trade, that
- 2:47is not how the room is set up.
- 2:48Take CoreWeave, a firm that buys Nvidia’s processors
- 2:52and rents out the computing power. In January
- 2:54twenty twenty-six, Nvidia paid cash to buy a
- 2:57block of CoreWeave’s shares. Then, under an agreement
- 3:00signed in September twenty twenty-five, Nvidia legally promised
- 3:04that it would buy whatever computing capacity CoreWeave
- 3:07reserves and cannot sell to anyone else, all
- 3:10the way through April twenty thirty-two. So Nvidia
- 3:13is CoreWeave’s supplier, one of its shareholders, and
- 3:16the buyer of whatever capacity it cannot rent
- 3:19out. Calling a company your customer when you
- 3:21own its shares and have contractually agreed to
- 3:23buy back whatever it fails to sell is
- 3:26a rather charming use of the English language.
- 3:28And that is only one room in the
- 3:30house. If you add up what Nvidia’s own
- 3:32filings say it has promised on behalf of
- 3:34its buyers, the total comes to about a
- 3:36hundred and sixty-five billion dollars. To be fair
- 3:40and exact, that figure is not money already
- 3:42lost and it is not corporate debt; it
- 3:45is the most those promises could cost. Those
- 3:47commitments are made of three very different things:
- 3:50promises to buy computing capacity its partners cannot
- 3:53sell, leases Nvidia has taken on data centers
- 3:56meant for others, and guarantees to pay landlords
- 3:59if a tenant defaults. Only that third bucket
- 4:02waits for someone to default; the first two
- 4:04fall due on their own schedule. The largest
- 4:07single piece is a guarantee signed in August
- 4:09twenty twenty-six on a cluster of data centers
- 4:12in Ohio being built for OpenAI on twenty-year
- 4:15leases, where Nvidia has promised to pay the
- 4:17rent if the tenant stops paying. On top
- 4:19of that, Nvidia holds about a hundred billion
- 4:22dollars of shares in other companies, and its
- 4:24unpaid customer bills have been stretching out for
- 4:27months. In fact, Nvidia’s finance chief stated in
- 4:30August that the artificial intelligence labs backed by
- 4:33Nvidia’s own balance sheet are expected to bring
- 4:35in roughly a quarter of its business next
- 4:38year. The cash changing hands is real, but
- 4:40so is the loop. When a seller has
- 4:42to stand behind its own buyers, it does
- 4:45not get to act surprised when a wobble
- 4:47on one side of the table shakes the
- 4:48other. Which brings us directly to the next
- 4:51question: can those customer companies actually afford to
- 4:54pay their own debts? When you look at
- 4:56the specialist firms doing the borrowing, the accounting
- 4:59looks remarkably thin. In the three months to
- 5:01September twenty twenty-six, CoreWeave owed about four times
- 5:05as much in interest as its underlying business
- 5:07generated in operating profit, and it finished the
- 5:10quarter with a loss. It owes tens of
- 5:12billions of dollars, and some of that debt
- 5:15is borrowed directly against the chips themselves. On
- 5:18November fifth, twenty twenty-six, some of CoreWeave’s bonds
- 5:21changed hands for as little as about seventy-five
- 5:24cents on the dollar. That means an investor
- 5:26holding a legal promise to receive a dollar
- 5:28in twenty thirty-two was willing to hand it
- 5:30to a stranger for three quarters of a
- 5:32dollar just to be rid of the paperwork.
- 5:35That seventy-five-cent price is an open-market trading price,
- 5:38a gauge of the lenders' anxiety rather than
- 5:40the borrowing rate the company locked in, but
- 5:43the anxiety is unmistakable. And it is not
- 5:46only CoreWeave: lenders are asking the biggest spenders
- 5:49for more too. Vanguard even issued a warning
- 5:52in September twenty twenty-six, reminding everyday savers that
- 5:55as these tech firms borrow, that corporate debt
- 5:58quietly slips into the broad, conservative bond funds
- 6:01that people like Carol rely on to preserve
- 6:03their savings. So if the weaker links are
- 6:06already feeling the strain, what would it actually
- 6:08take to bring this boom down? It would
- 6:10take a simple change of mind inside a
- 6:13very small number of boardrooms. The real weight
- 6:15in this market sits with four companies: Microsoft,
- 6:19Alphabet, Amazon, and Meta. In twenty twenty-six, those
- 6:23four firms planned to spend more than seven
- 6:25hundred billion dollars between them on buildings and
- 6:28equipment, much of it for artificial intelligence. That
- 6:31river of spending is where Nvidia’s sales come
- 6:34from, it is the reason the smaller cloud
- 6:36providers can borrow, and it is the foundation
- 6:38Wall Street is leaning on. But pouring that
- 6:41much money into concrete and silicon is beginning
- 6:44to stretch even the largest corporate treasuries. Over
- 6:47the first nine months of twenty twenty-six, Alphabet
- 6:49spent slightly more on buildings and equipment than
- 6:52its whole business brought in in cash, and
- 6:55it is not alone; some of them are
- 6:56now spending more cash than their businesses bring
- 6:59in. The comfortable assumption is that all this
- 7:01rests on something solid, because people hear the
- 7:04phrase order book and picture a vault of
- 7:07guaranteed income. But if you open Nvidia’s own
- 7:10annual filing and read the plain print, it
- 7:12says its customers can generally cancel, change, or
- 7:15delay their orders with little notice and without
- 7:18penalty. Those are not ironclad sales; they are
- 7:21informal reservations. So the whole structure depends on
- 7:24four companies choosing to keep spending at this
- 7:27historic pace, quarter after quarter, on orders they
- 7:30can call off for free. To see what
- 7:32happens if they change their minds, one major
- 7:34bank ran the arithmetic in November twenty twenty-six:
- 7:38if those big spenders reduce their capital budgets
- 7:40by twenty percent in twenty twenty-seven, the companies
- 7:43that make artificial intelligence chips would lose more
- 7:46than a third of their profits, and possibly
- 7:48more. That twenty percent cut is a what-if
- 7:51scenario, not a forecast that it will happen,
- 7:54but it is the clearest way this entire
- 7:56market could unwind. When hundreds of billions of
- 7:59dollars of revenue depend on four chief executives
- 8:02not calling off orders they can cancel at
- 8:04no cost, you are leaning entirely on human
- 8:07mood. And a spending cut is not the
- 8:09only way the boom could stumble, especially now
- 8:12that Nvidia’s chief executive is gone. Back in
- 8:15July twenty twenty-six, officials at the Federal Reserve
- 8:18explicitly warned that if artificial intelligence fails to
- 8:21live up to corporate expectations, the disappointment could
- 8:24knock share prices down broadly and cause people
- 8:27to cut back on their spending. To be
- 8:29fair, Jensen Huang’s exit was not the sole
- 8:32cause of the stock’s five-week slide; about seven
- 8:35tenths of that thirty percent drop had already
- 8:37happened before the email was published on November
- 8:40eighth. The market was already nervous about something
- 8:43wider than one person. What a change at
- 8:45the top can alter is the financing arrangements
- 8:48Nvidia uses to keep its customers buying. The
- 8:50contracts already signed bind the company whoever sits
- 8:54in the corner office; the Ohio guarantee, for
- 8:56instance, belongs to Nvidia as a corporation. But
- 9:00offering new money to buyers is an ongoing
- 9:02choice, and on the day of the resignation,
- 9:04investors openly worried that a new chief executive
- 9:07might be far less willing to bankroll customer
- 9:10purchases. Analysts at one bank even raised the
- 9:13possibility that about fifty billion dollars of Nvidia's
- 9:16funding for other companies could be put on
- 9:18hold. No pause has been officially announced, and
- 9:21nobody can yet know how a new leader
- 9:23will manage those checkbooks. My view of Wall
- 9:26Street trying to predict that transition is that
- 9:28they are guessing out loud in very confident
- 9:31voices, because admitting they do not know is
- 9:33the one thing they are not allowed to
- 9:35do. Whenever an essential supplier loses thirty percent
- 9:39of its value while insisting its technology will
- 9:41change civilization, people reach for the history books,
- 9:44and the chapter they open is March two
- 9:46thousand. Back then, the undisputed centerpiece of the
- 9:50digital frontier was Cisco Systems. On March twenty-seventh,
- 9:54two thousand, Cisco became the most valuable public
- 9:57corporation on earth. It was not an empty
- 10:00dot-com startup; it sold the equipment the internet
- 10:02was being built with, and it made billions
- 10:04of dollars a year in real profit. But
- 10:07when its customers stopped building, Cisco’s business stalled.
- 10:10In the following year, it was writing off
- 10:12billions of dollars of equipment it could not
- 10:15sell and reporting a loss, even while cash
- 10:17kept flowing through the doors. The expectation every
- 10:20investor is taught is that if a company
- 10:22is real, essential, and profitable, and if it
- 10:25survives the bust, its share price will inevitably
- 10:28recover; you only have to wait. Cisco survived,
- 10:32it remained an indispensable supplier, and the internet
- 10:35turned out to be everything its believers said
- 10:37it would be. But its share price did
- 10:39not close above that March two thousand high
- 10:41until December tenth, twenty twenty-five, more than twenty-five
- 10:45years later. A shareholder who was right about
- 10:47the technology, right about the company, and right
- 10:50about its survival still had to wait a
- 10:52quarter of a century just to see the
- 10:54share price they paid again. And Cisco was
- 10:57not the only hardware giant that learned that
- 10:59lesson. Lucent Technologies lent its own customers the
- 11:02money to buy its equipment, promising loans and
- 11:05guarantees so they could keep buying its products,
- 11:07and when those customers could not pay, Lucent
- 11:10charged nearly two billion dollars against that bad
- 11:13lending in a single year; its shares lost
- 11:15nearly all their value. When that broader market
- 11:18cleared, the Standard and Poor’s five hundred had
- 11:21lost about half its value and took seven
- 11:23years to get back to its peak. None
- 11:25of that proves Nvidia will walk the same
- 11:27path, but it demonstrates that a technology can
- 11:30triumph while the people who bought the hardware
- 11:32supplier at peak prices get taken apart. That
- 11:35was their stock twenty-five years ago. The question
- 11:38that matters to Carol is how much of
- 11:40this current machine sits inside her own four-oh-one-k
- 11:43without her ever choosing it. At Vanguard, more
- 11:46than four fifths of savers in workplace plans
- 11:48keep their money in a target-date fund. In
- 11:51an illustration built from Vanguard’s published asset allocation
- 11:54and portfolio holdings from late September twenty twenty-six,
- 11:57a typical retirement account holds about two to
- 12:00four cents of Nvidia in every single dollar.
- 12:03Older workers like Carol sit near the low
- 12:05end of that range, while younger workers like
- 12:07Ben sit near four cents, because target-date funds
- 12:10automatically reduce stock exposure as a worker nears
- 12:13retirement. That means Nvidia's thirty percent drop over
- 12:17the last five weeks, taken entirely by itself,
- 12:20has cost Carol about eight hundred and fifty
- 12:22dollars of her hundred and seven thousand. That
- 12:24eight hundred and fifty dollars sounds entirely manageable,
- 12:27but stopping there is a mistake, because the
- 12:30real exposure is not Nvidia alone. An index
- 12:33fund tracking the Standard and Poor’s five hundred
- 12:36is sold as the ultimate sensible investment because
- 12:38it spreads your savings across five hundred distinct
- 12:41American companies; that is the promise printed on
- 12:44the brochure. But by November eleventh, twenty twenty-six,
- 12:47the ten largest corporations in that index made
- 12:50up about thirty-eight cents of every single dollar
- 12:53in the fund. At the peak of the
- 12:55dot-com bubble in March two thousand, the top
- 12:57ten companies were only about twenty-seven cents of
- 13:00each dollar. Today, eight companies closely tied to
- 13:03the artificial intelligence build-out represent more than a
- 13:06quarter of the entire fund. In fact, by
- 13:09one major fund company’s own calculation, the Standard
- 13:12and Poor’s five hundred now spreads money about
- 13:14as widely as holding fifty equally sized companies.
- 13:18For Carol, the ten largest American corporations represent
- 13:21about an eighth of her entire account, bonds
- 13:24included, while for Ben they make up about
- 13:26a fifth. And neither of them can instruct
- 13:28the fund to sell its Nvidia; you own
- 13:31the mutual fund, not the companies inside it.
- 13:34It is a peculiar kind of diversification that
- 13:36promises you five hundred companies and hands you
- 13:38fifty, with nearly four dollars in ten riding
- 13:41on ten corporate boards. What that concentration does
- 13:44to your savings depends heavily on how many
- 13:47years you have left on the job, which
- 13:49is the deeply unfair part of retirement finance.
- 13:52The funds are built to soften blows by
- 13:54shifting into bonds, keeping about nine tenths of
- 13:57Ben's money in stocks at age thirty and
- 13:59about six tenths of Carol's at age sixty.
- 14:02But that mechanical shift does not make an
- 14:04account immune to a broad collapse. During the
- 14:07financial crisis of two thousand and eight, the
- 14:09average workplace account held all the way through
- 14:12lost about a third of its value in
- 14:14a single twelve-month stretch. If a similar broad
- 14:17contraction hit the equity markets tomorrow, that average
- 14:20fall would take about thirty-seven thousand dollars from
- 14:23an account the size of Carol's. That is
- 14:25an illustration of the size, not a forecast
- 14:28for her fund. We saw this exact generational
- 14:31divide during the dot-com bust. Savers who were
- 14:34in their sixties when that bubble burst had,
- 14:36on average, about a fifth less in their
- 14:38accounts three years later, and even after contributing
- 14:41fresh money from their paychecks for an extra
- 14:43year, they were still short of where they
- 14:45started four years into the slump. Meanwhile, savers
- 14:49who were in their twenties saw their average
- 14:51balances grow through the crash, because the fresh
- 14:53contributions they poured in each month were large
- 14:56compared to the modest savings they had accumulated.
- 14:59That is Ben’s structural cushion: a market crash
- 15:02acts as a long-term sale for a thirty-year-old,
- 15:05provided he keeps his job and keeps buying.
- 15:07But for money already sitting in a Standard
- 15:09and Poor’s five hundred fund with nothing new
- 15:12added, an investor who held at the end
- 15:14of nineteen ninety-nine was not back to even
- 15:16until the end of two thousand and six.
- 15:19And the damage is far worse for anyone
- 15:21who has already retired and must sell shares
- 15:23each month to buy groceries, because those depressed
- 15:26shares are permanently gone from the ledger when
- 15:29the market finally recovers. I dislike the idea
- 15:32that a worker five years from retirement is
- 15:34told she was reckless, when all she did
- 15:36was enroll in the default fund her employer
- 15:39selected. Now, before anyone concludes that the entire
- 15:42enterprise is falling apart tomorrow, we should look
- 15:45fairly at Nvidia’s actual business, because the people
- 15:48arguing that the company is wildly profitable are
- 15:51describing a real balance sheet. In the six
- 15:53months to late July twenty twenty-six, Nvidia’s operations
- 15:57brought in approximately seventy-four billion dollars of pure
- 16:00cash. It holds considerably more in cash and
- 16:03bonds than it owes in debt. Furthermore, issuing
- 16:06a financial guarantee is not the same as
- 16:09writing a check today: the large Ohio commitment
- 16:11is not expected to begin phasing in until
- 16:14its twenty twenty-nine financial year. Those are genuine
- 16:17strengths, but a sound operating business is not
- 16:20the same thing as a safe share price;
- 16:22Cisco was generating billions in cash at its
- 16:24peak too. The uncomfortable detail is that every
- 16:27financial report Nvidia has published covers performance from
- 16:31before the chief executive resigned, with the next
- 16:33official scorecard not arriving until November seventeenth. Beneath
- 16:37the surface, customers are already taking longer to
- 16:40pay, with the average collection window stretching from
- 16:43forty-five days to sixty in a single quarter.
- 16:46And two of Nvidia's largest commitments require no
- 16:49customer default to fall due: it has pledged
- 16:51to buy about thirty-six billion dollars of computing
- 16:54capacity from its partners over roughly six years
- 16:57if they cannot rent it out, and it
- 16:59has signed leases on data centers that it
- 17:01must pay for if nobody else takes them
- 17:03over. A healthy income statement is comforting, but
- 17:06it does not protect you when your future
- 17:08sales depend on propping up the very people
- 17:10who buy from you. Nor does the broader
- 17:12market’s apparent calm prove that the danger has
- 17:15passed. When you look at headline index numbers,
- 17:18the system appears remarkably steady. Even as Nvidia
- 17:21dropped thirty percent over five weeks, the broader
- 17:24Standard and Poor’s five hundred ended November twelfth
- 17:27just three percent below its all-time record, sitting
- 17:30comfortably in positive territory for the year, while
- 17:32corporate profits grew across every sector. Yet that
- 17:36calm cracked in plain sight on November fifth,
- 17:39when the Standard and Poor’s five hundred fell
- 17:41more than four percent in a single day.
- 17:43Semiconductor shares as a group ended November eleventh
- 17:46down more than a quarter from their June
- 17:48peak. The index has held together so far,
- 17:51but yesterday’s stability is no guarantee that the
- 17:53rest of the market stays contained. Ever since
- 17:56two thousand and eight, many savers hold an
- 17:58instinctive belief that if things get bad enough,
- 18:01a government bailout will step in to make
- 18:03them whole, because a rescue was sold back
- 18:05then as protecting the broader economy. But under
- 18:08rules adopted by the Federal Reserve in twenty
- 18:11fifteen, the central bank is explicitly barred from
- 18:14lending to rescue one particular failing company. The
- 18:17White House’s own adviser on artificial intelligence stated
- 18:21publicly a year ago that there will be
- 18:23no federal bailout for artificial intelligence corporations. Even
- 18:27when bailouts do arrive, the word means something
- 18:29very different from protecting shareholders. When the federal
- 18:32government rescued General Motors, the factories kept running
- 18:36and the business survived, but the old company’s
- 18:38shares were canceled in March twenty eleven with
- 18:41nothing paid to the people who owned them.
- 18:43A rescue protects the business and its operations;
- 18:46the shareholders are the ones it is rescued
- 18:48from. And common shares are exactly what sit
- 18:51inside Carol’s four-oh-one-k. So how real is the
- 18:55worst case, has the boom broken, and how
- 18:58much of your four-oh-one-k has actually gone with
- 19:00it? The answer, based on the documented evidence,
- 19:03is not so far, but the risk is
- 19:05very real. It has not broken yet because
- 19:08while Nvidia fell about thirty percent, the broader
- 19:11Standard and Poor’s five hundred dropped only about
- 19:14three percent, and an age-balanced fund like Carol’s
- 19:16lost less than one percent between Nvidia’s peak
- 19:19and the day before the announcement. Ben’s fund,
- 19:22with its heavy stock weighting, lost only a
- 19:24little more. In funds like theirs, Nvidia itself
- 19:27makes up only about two to four cents
- 19:29of every dollar, which is too small to
- 19:32sink an account on its own. The deciding
- 19:34reason for the danger lies elsewhere: the risk
- 19:37is that Nvidia, the cloud operators it finances,
- 19:40the bondholders funding those clouds, and the ten
- 19:42corporate giants that make up thirty-eight cents of
- 19:45every index dollar are all leaning on the
- 19:47exact same decision. That decision is whether a
- 19:50handful of tech giants keep spending more than
- 19:53seven hundred billion dollars a year on buildings
- 19:55and equipment, much of it for artificial intelligence.
- 19:58If they reduce those budgets, the consequences move
- 20:01the entire top of the market at once;
- 20:04as the bank’s scenario showed, a twenty percent
- 20:06cut in spending takes more than a third
- 20:08off chipmaker profits. That is the clearest path
- 20:11to the worst case, alongside a wider loss
- 20:14of faith in artificial intelligence earnings or a
- 20:17pullback in financing under Nvidia’s new management. What
- 20:20will show which way this goes is not
- 20:22the daily gossip, but the numbers: Nvidia’s results
- 20:25on November seventeenth, and the capital expenditure budgets
- 20:28published by Microsoft, Alphabet, Amazon, and Meta over
- 20:32the coming quarters. If those budgets do get
- 20:35cut, the cost will not be shared equally.
- 20:37A retirement saver does not face the same
- 20:40clock at thirty as they do at sixty.
- 20:42If the artificial intelligence build-out falters, history shows
- 20:46that an indispensable supplier like Cisco can spend
- 20:49twenty-five years waiting to see its old share
- 20:51price again, while an equity index can spend
- 20:54seven years getting back to even. Ben has
- 20:57thirty-five years of paychecks ahead of him; he
- 20:59will keep buying through any downturn, picking up
- 21:02shares at a discount. Carol has five years
- 21:04until she plans to retire, with an account
- 21:07holding a hundred and seven thousand dollars that
- 21:09she will soon need to convert into monthly
- 21:11living expenses. And there is no administrative rule
- 21:14on the books that will bail out the
- 21:16shares inside her account. Nothing has broken yet.
- 21:19I don't think that is the same as
- 21:21safe: it only means the few companies everybody
- 21:24is leaning on are still spending. Thanks for
- 21:26watching! Please subscribe, it's free, it genuinely helps,
- 21:28and watch this next, it's right there, okay,
- 21:29thank you, bye!
About this transcript
This page contains the full transcript of Nvidia’s Troubles Could Become America’s Crisis by dummy finance, generated from the public captions YouTube serves with the video. The transcript has 3,883 words across 486 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.