The Nasdaq hit a record, Nvidia neared $6 trillion, and software stocks soared. All while the 10-year paid 5.3%. One of these markets is wrong.
Published 7 October 2026 · 06:00 GMT
Wall Street is running with the bulls and the bond market at the same time, and the contradiction is the story. The Nasdaq closed Monday at a record 27,477.31, up 1.05 percent, and on Tuesday the S&P 500 notched its first record close since August 13, with the Nasdaq and the Magnificent Seven ETF booking second straight records. Nvidia, the emblem of the AI trade, rose 2.1 percent Monday to a record close valuing the company at $5.76 trillion, about 4 percent shy of the first $6 trillion market capitalization in history. All of this happened with the 10-year Treasury yielding 5.31 percent at Monday’s close, its highest settlement since 2002. As Deutsche Bank put it, bonds and stocks are pricing two irreconcilable worlds, and one of them will have to yield.
The AI trade’s latest leg has specific fuel. Nvidia’s September 28 announcement of a record $150 billion buyback increase, the largest in American corporate history, reframed the company as a cash machine as well as a growth story; chief executive Jensen Huang called AI a once-in-a-generation platform shift and said chip shipments will double in 2027. The stock is up 28 percent this year, adding $1.2 trillion of market value. The Magnificent Seven together closed Tuesday at a combined record of about $24.95 trillion, just shy of $25 trillion. When seven companies are worth a quarter of a hundred trillion dollars, the index is no longer diversified; it is a conviction.
When seven companies are worth a quarter of a hundred trillion dollars, the index is no longer diversified; it is a conviction.
The more surprising leg is software. The S&P 500 software and services index rose 1.3 percent Tuesday to its highest since November 2025, capping its biggest quarterly jump since the second quarter of 2020. Expected 2026 earnings growth for the group has climbed to 20.6 percent from 13.8 percent at the end of March. Marvell Technology jumped 5.8 percent to $287.01 after an investor day set a $70-90 billion revenue target for fiscal 2031, against $8.2 billion expected this year, and the stock is up 250 percent in 2026. Cisco added 4.5 percent. As Adam Turnquist of LPL put it, AI has been more of an enabler for software companies than a disruptor: the feared replacement of software by AI has inverted into AI as software’s best salesman.
The exception proves the rule’s selectivity. Seagate is down 14 percent cumulatively since October 1, first on Toshiba doubling hard-drive capacity plans for fiscal 2027, then on a bidding war with Toshiba for TDK’s magnetic-heads unit. In a market this narrow, stock-picking still matters; the index record is an average that hides violent dispersion underneath.
The contradiction sits in the discount rate. The 10-year closed Monday at 5.311 percent after an intraday peak of 5.347 percent, the highest since April 2002, before cooling to 5.270 percent Tuesday. At those yields, the risk-free asset pays what equities once promised, and every growth stock must justify its multiple against a hurdle that keeps rising. That the market keeps clearing the hurdle is either evidence that AI earnings are real enough to outrun 5.3 percent, or evidence that the hurdle is about to move.
Keith Lerner of Truist captured the market’s self-soothing logic: tech is almost seen as defensive in some ways, the fortress balance sheets of the mega-caps functioning as the new safety trade. Dan Russo of Potomac was blunter: it is only the fortress-like balance sheets at the large-cap end of the spectrum propping the market up. Both descriptions are true, and both are warnings. Defensive tech is an oxymoron that works until the cycle turns, and a market propped by seven balance sheets is a market with seven single points of failure.
The bond selloff’s global reach adds an uneasy coda. Japan’s 30-year yield hit 4.235 percent on Monday, an all-time record since the bond’s 1999 debut, and the 10-year sits at its highest since August 1996. Bank of Japan Deputy Governor Shinichi Uchida called AI a big positive demand shock for the economy while warning of correction risk if profits do not materialize, a central banker blessing the boom and hedging it in the same sentence. If even Japan, the land of permanently low rates, now prices long money above 4 percent, the era of cheap capital is over everywhere, and equities are partying on its grave.
Listen to the analysts and a new consensus is forming: AI stopped being software’s executioner and became its sales force. William Kerwin of Morningstar called Marvell’s $70-90 billion revenue target a strong endorsement that huge AI spending continues for five more years. Jim Awad of Clearstead argued Nvidia is attractive on both a growth and a value basis, a remarkable sentence to write about a $5.76 trillion company, and one that captures how the valuation debate has shifted: at 28 percent gains this year on real cash flows and a record buyback, the bulls no longer need to invoke the future; they can point to the present.
The numbers underneath support the chorus, up to a point. Software’s expected 2026 earnings growth of 20.6 percent, up from 13.8 percent in March, is the kind of revision that justifies multiples, and the group’s biggest quarterly jump since 2020 suggests the AI uplift is broadening beyond the chipmakers. But the concentration remains the shadow over the story: the Magnificent Seven at nearly $25 trillion means seven earnings calls decide the market’s fate, and a single disappointment, a capex cut, a margin warning, reprices the conviction trade all at once.
The referee arrives next week. Third-quarter earnings season kicks off with the big banks, JPMorgan reporting October 13, and S&P 500 earnings are expected to grow more than 30 percent year on year, largely AI-driven. That expectation is the entire bull case in one number: if the banks confirm credit is clean and the hyperscalers confirm AI spending continues, the records validate themselves. If either wobbles, the Deutsche Bank irreconcilability resolves the other way, fast.
Art Hogan of B. Riley described the market’s mood with trader’s honesty: with the data calendar light and earnings a week away, investors are groping for anything positive to latch onto. Groping is not a strategy, but it has been a profitable one. The question for the next two weeks is whether earnings turn the groping into grip.
The stakes are larger than portfolio statements. A $5.76 trillion Nvidia and a $25 trillion Magnificent Seven mean the AI buildout is now systemically important: pension funds, index trackers, and sovereign wealth funds all own the trade whether they chose it or not. If AI earnings deliver, the productivity boom reprices everything upward and the 5.3 percent yield becomes a footnote. If they do not, the correction will not be a tech event but a market event, transmitted through every index fund on earth.
For now, the market has made its bet: that intelligence is the new oil, that the buildout pays for itself, and that 5.3 percent is a price worth paying for a ticket to the platform shift. It is the biggest concentrated bet in financial history. Like all such bets, it will look inevitable in retrospect, right up until it doesn’t.
From the Western markets lens, the AI boom is the last great growth story that fits inside the old valuation math, and that is precisely the bull case. Western analysts argue the Magnificent Seven earn their multiples the old-fashioned way, with cash: $150 billion buybacks, 20.6 percent software earnings growth, fortress balance sheets that make 5.3 percent an annoyance rather than a threat. In this telling, tech as defensive is not an oxymoron but an upgrade of the category, the new consumer staples for an economy that runs on intelligence.
The Western bear case is concentration, not valuation. Seven companies at $25 trillion means the market’s fate is a committee decision by seven management teams, and Western risk managers are quietly uneasy about an index that is no longer an index. The honest Western summary: the trade is fundamentally sound and structurally fragile at the same time, a combination that has historically resolved through volatility rather than through calm.
From Beijing, the American AI boom is watched the way a chess player watches the opponent’s queen: with respect, and with plans. The Eastern lens reads $5.76 trillion of Nvidia as both an American triumph and an American vulnerability, a single company on which the entire Western productivity narrative now depends. Chinese analysts note the symmetry with open satisfaction: Washington restricted China’s access to advanced chips to slow it down, and the result is an American market priced for an AI future whose hardware bottleneck runs through a handful of companies, one island’s fabs, and one company’s GPUs.
Moscow’s read is more sardonic. Every trillion of American tech valuation is a trillion not spent on the material economy, and Russia, selling oil at $101 into the same inflation the Fed fights at 5.3 percent, is content to watch financial engineering compound. The Eastern consensus is that the AI boom is real technology wrapped in unreal expectations, and that the reckoning, when it comes, will be priced in the one currency neither side controls: time.
From the Global South, the $25 trillion Magnificent Seven is a number from another planet, and the distance is the point. Southern analysts do the development math with dry humor: the combined value of seven American companies exceeds the GDP of every African economy put together several times over, and the AI future being priced in New York will arrive in Lagos and Dhaka as a subscription fee. The South does not resent the boom; it recognizes the pattern, every technological revolution prices its gains in the North first and sends the bill South later.
The substantive Southern question is about the buildout’s material footprint. Data centers need power, chips need minerals, and the minerals come from Southern ground: the AI boom’s supply chain runs through Congolese cobalt, Chilean copper, and Indonesian nickel. The South’s leverage is geological, and Southern strategists are increasingly explicit that the next decade’s bargain will be struck on those terms. Wall Street prices intelligence; the South owns the atoms intelligence runs on. That negotiation has barely begun.
Nvidia closed Monday at a $5.76 trillion market valuation, up 28% this year with $1.2 trillion added, about 4% short of becoming the first $6 trillion company in history. Its $150 billion buyback increase in September was the largest in US corporate history.
The market is betting AI earnings will outrun the discount rate: software stocks hit their highest since November 2025 on 20.6% expected 2026 earnings growth, and the Magnificent Seven’s fortress balance sheets are treated as defensive. Deutsche Bank’s counter is that bonds and stocks price irreconcilable worlds, and one must give.
Marvell jumped 5.8% to $287.01 after telling investors it targets $70-90 billion of revenue in fiscal 2031 versus $8.2 billion expected this year, a bet that AI infrastructure spending continues for five more years. The stock is up 250% in 2026.
Seagate is down 14% cumulatively since October 1, not in a single day: first on Toshiba’s plan to double hard-drive capacity in fiscal 2027, then on a bidding war between Seagate and Toshiba for TDK’s magnetic-heads unit. It shows the index record hides sharp dispersion underneath.
Third-quarter earnings starting next week, with JPMorgan on October 13 and S&P earnings expected up more than 30%. If banks confirm clean credit and hyperscalers confirm continued AI spending, the records validate. If either wobbles, the bond market’s verdict arrives fast.
