On Election Day, the bond market has already voted. What Oracle's borrowing costs tell us about the next phase of the AI boom.
KEY TAKEAWAYS
AI stocks entered a deeper correction as investors reassessed financing risks. The Nasdaq 100 declined 1.4% on Monday, closing 10.4% below its October 6 record, while the semiconductor index fell 12% over three sessions.
Oracle announced a $20 billion capital raise just five days after outlining its financing limits through fiscal 2030. The offering includes $10 billion in common stock priced at an 11% discount and $10 billion in senior notes carrying interest rates as high as 8.65%.
Oracle's borrowing costs reflect a significantly higher credit risk premium. With the 10-year Treasury yielding approximately 4.91% and Oracle rated BBB- by S&P, its highest-coupon notes carry a substantial premium over government debt.
Marvell announced a $115 billion all-stock acquisition of Lumentum, offering 4.5 Marvell shares for each Lumentum share. Despite the implied 23.9% acquisition premium, Marvell shares declined 12.7% and Lumentum shares fell 18.6% following the announcement.
Upcoming earnings and economic reports will provide additional tests for AI valuations and financing conditions. AMD reports Tuesday, Qualcomm and Arm report Wednesday, and Friday's employment report could influence expectations for the Federal Reserve's December policy decision.
MY HOT TAKES
The AI investment narrative is shifting from opportunity to affordability. Investors spent three years asking how large the AI market could become, but they are finally asking the more consequential question: who will finance it, and at what cost?
Oracle's $20 billion raise is more revealing than its official financing guidance. Exhausting an entire four-year capital-market funding allowance in five days raises uncomfortable questions about what happens when the company needs additional capital.
Marvell's acquisition demonstrates the dangers of using inflated stock valuations as acquisition currency. When both the acquirer and its target decline sharply after an all-stock deal is announced, investors are questioning not just the transaction but the value of the paper financing it.
Today's AI financing environment bears uncomfortable similarities to the late-1990s telecom boom. The technology and underlying demand are different, but the dependence on equity valuations, aggressive acquisitions, and increasingly expensive capital looks remarkably familiar.
Investors should be scrutinizing balance sheets and financing structures as carefully as growth forecasts. Companies generating sufficient internal cash flow have a fundamentally different risk profile from those relying on continuous equity issuance or expensive borrowing to sustain expansion.
You can quote me: "Paper is the cheapest money in the world–right up until the day the market stops accepting it at face value."
Paper trade. It's Election Day, and I already know how my morning went—because it went the way every Election Day morning goes. First espresso, local TV news humming in the background, and EVERY channel running the same segment. Reporters in puffy jackets stand outside school gymnasiums, pointing at lines that haven't formed yet and interviewing the one voter who showed up before sunrise with a thermos (There is always a guy with a thermos–at least in NYC 🤣). The weather gets ninety seconds instead of four minutes, the subway delays get exiled to the crawl at the bottom of the screen, and everyone–anchors, pundits, the guy with the thermos–is waiting for tonight. Waiting for the polls to close, for the counting to start, for somebody on a cable set to say "too close to call." But while the whole country waits to vote, a different electorate already cast its ballots–yesterday–and nobody put it on the morning news. 🫤 That’s right–the bond market voted. And the bond market doesn't wait for the polls to close. It votes EVERY day, it votes with money, and yesterday it voted on Oracle at 8.65%.
Monday was the day the AI trade stopped asking the question it has been asking for three years–how BIG is this opportunity?--and started asking a much older and much less fun question. Who is paying for it? And with what? The Nasdaq 100 fell 1.4% and closed 10.4% below its October 6 record, which puts it officially in correction territory (a correction, you will recall, is a decline of 10% or more from a recent high). The semiconductor index has now dropped 12% in three sessions. That isn't a rotation. That's a repricing. And the catalyst wasn't a missed earnings number or a hot inflation print–it was two financing announcements that told investors, in plain English, that cash flow alone is no longer paying the bills.
Start with Oracle, because Oracle's timing was–how shall I put this–er, exquisite. Last Wednesday, at its Investor Day, Oracle told the world it would limit additional capital-market funding to $20 billion for the entire investment cycle running through fiscal 2030. Four years. Twenty billion. A ceiling, a promise, a line in the sand. On Monday–five days later–Oracle filed to raise exactly $20 billion, split between $10 billion of new common stock and $10 billion of senior notes. The whole four-year allowance, spent in a week–a WEEK! To be fair, Oracle didn't technically break its promise; it raised right up to the ceiling, not through it. But when you exhaust your four-year budget before the following Monday, the market is entitled to wonder what happens in year two. And the terms tell you exactly how the market feels. The new shares were priced at an 11% discount. That means new buyers paid meaningfully less than the stock was worth the day before, so existing holders got diluted AND marked down at the same time, a two-for-one special nobody ordered. The notes carry interest rates from 7% to 8.65%. Let me put that in perspective. The 10-year Treasury yields about 4.91%. Oracle, a household name and a fixture of corporate America for decades, is paying nearly four full percentage points more than that to borrow. That isn't a technology company's cost of capital. That's a credit story. And it fits, because S&P cut Oracle to BBB- in July, one notch above junk. The stock fell 8.8% on Monday and now sits more than 60% below its September 2025 peak.
Then there's Marvell. On Monday, Marvell announced it would acquire Lumentum, the maker of the lasers and optical components that move data between AI chips, in an all-stock deal carrying an implied enterprise value of $115 billion (enterprise value is the price tag on the whole company, debt included, not just the shares). Lumentum holders get 4.5 Marvell shares for every share they own, roughly a 23.9% premium, and when the dust settles they'll own about a third of the combined company. Why all stock? Because Marvell had $3.93 billion of cash on its balance sheet as of August 1. You don't buy a $115 billion company with $4 billion in the checking account. You buy it with your own paper. And remember the scale here. Less than a year ago, in December 2025, Marvell agreed to buy the photonics startup Celestial AI for $3.25 billion in cash and stock. Going from a $3 billion deal to a $115 billion deal isn't the next step up the staircase, it's taking the express elevator to the roof. And the market delivered the most honest review I've seen all year: Marvell fell 12.7%, and Lumentum fell 18.6%. Read that again. The TARGET fell. Normally the company being acquired rallies toward the deal price–that's the whole point of a premium. When the target sinks right alongside the buyer, the market is telling you it doesn't think much of the currency being used to pay. Lumentum shareholders are being paid in Marvell shares, so every dollar Marvell loses, Lumentum loses too. They're chained together.
I've seen this movie. I had a front-row seat to it, actually–I was at Lucent during the telecom boom, doing tech M&A in Murray Hill when an entire industry discovered that its stock was worth more as currency than as ownership. In January 1999, Lucent agreed to buy Ascend Communications for about $20 billion–in stock–with each Ascend share converting into 0.825 Lucent shares. At the time, industry sources told reporters that Lucent had been waiting for its share price to reach the right level before making the offer. Think about that sentence. The timing was tied to the STOCK PRICE. When your shares are your currency, your share price stops being a scorecard and becomes the business plan. And it worked beautifully…until the stock fell. Then the currency was worth less, the acquisitions were worth less, and the customers–the CLECs (competitive local exchange carriers, the upstart phone companies of the late 1990s), many of whom had bought all that equipment with borrowed money–stopped being able to pay. And here's the detail that still gets me. On the day the Ascend deal was announced, Lucent's own stock fell nearly 4%. The buyer got punished then, too, which is a polite little warning shot that almost nobody heeded. The whole thing unwound in the same order it was built, just faster.
Aside 😉: I worked on the Ascend deal directly and while stock price was an important factor (as largely reported in the media), the deal was largely driven by strategy and Lucent’s immediate necessity to fill a gap in its product offering portfolio. More, aside 😉😉: The Marvell deal and, in fact, a lot of the more recent tech deals are starting to look strikingly similar to deals I worked on back in the late 1990s–just saying.
To be clear, I'm not going to tell you 2026 is 1999 with better Wi-Fi. It isn't, and pretending otherwise would be lazy. The biggest buyers of AI infrastructure today aren’t debt-soaked startups. Many of them are among the most profitable companies on the planet. The demand is real. The products work. But here's what rhymes, and it rhymes loudly. Back then, nobody panicked while the boom was being funded. They panicked when the funding got expensive. On Monday, the funding got expensive. An 11% discount on new stock. Coupons north of 8%. An all-stock megadeal where the target sank. That's the market quietly raising the price of admission, and it did it on the same day the Nasdaq 100 crossed into correction.
So, what should an investor do with this? I'm not going to tell you what to buy or sell, but I'll tell you the questions I'd be asking about every AI-adjacent name in my portfolio this week. Is this company funding its growth from the cash it generates, or from the paper it prints? If it's printing paper, who absorbs the dilution–and is it me? What is the bond market charging it to borrow today versus a year ago? And if it's buying companies with its own stock, is that stock rich enough to make the deal a bargain, or cheap enough to make it a fire sale? None of those are trick questions, but I'd argue the answers to every one of them changed between Friday's close and Monday's. The next tests arrive fast. AMD reports after the bell tonight, while most of us are watching election returns. Qualcomm and Arm report Wednesday. And Friday's jobs report will help decide whether the Fed tightens in December, which would make all of that paper…well, more expensive still.
Here's the part nobody wants to hear. Paper is the cheapest money in the world–right up until the day the market stops accepting it at face value. Monday wasn't that day. But it was the first day in a long time that the market checked the bill twice.
Tonight the country will watch the returns roll in, precinct by precinct, and somebody on a cable set will eventually say "too close to call." I'll be watching too, with something a little stronger than espresso, I suspect. But I already know the result of one race. The bond market voted yesterday. Turnout was heavy, the count was fast, and nobody needed a recount–the winning number was 8.65%. And I'd bet the guy with the thermos didn't even notice. 😉
YESTERDAY’S MARKETS
Stocks started November mostly lower as a three-day chip selloff pushed the Nasdaq 100 into correction territory. The Dow Jones Industrial Average gained 214 points, or 0.41%, to close at 51,983, while the S&P 500 fell 0.51%, the Nasdaq Composite slid 1.07% and the Russell 2000 slipped 0.21%. Oracle fell 8.8% after filing for a $20 billion stock-and-debt raise, while Marvell dropped 12.7% and Lumentum 18.6% after the two announced an all-stock merger. Oil settled sharply lower on hopes for progress in U.S.-Iran talks, with WTI down 7.36% to $96.46 and Brent down 7.63% to $104.77, while the 10-year Treasury yield fell 3.5 basis points to 4.89%.
NEXT UP
JOLTS Job Openings (September) may have increased to 7.164 million vacancies from 7.079 million.
Factory Orders (September) probably increased by 1.2% after climbing by 0.1% in the prior month.
Important earnings today: Yum! Brands, Apollo Global Management, Archer-Daniels-Midland, Keurig Dr Pepper, Zebra Technologies, Marathon Petroleum, Pfizer, ADT Inc, Fiserv, Energy Transfer, Susco, Booking Holdings, Block Inc, International Flavors and Fragrances, Duokingo, Mosaic, AMD, Skyworks Solutions, Super Micro Computer, and Amgen.