Everything About AI
NextFin News - The stock market rang in the end of election uncertainty with a broad rally on Wednesday, but the companies that powered this year's artificial-intelligence boom did not join the celebration - and the selloff deepened into Thursday. The Dow Jones Industrial Average jumped 1,569.40 points, or 3.03%, to 53,356.12, its best single session since April 9, 2025. The S&P 500 gained 1.58% to 7,688.62 and the Nasdaq Composite added 1.44% to 25,736.51. Yet the iShares Semiconductor ETF (SOXX) fell 3.14%, Nvidia dropped 3.17% to $200.13, Micron slid 7%, Marvell lost 7.08%, and Sandisk tumbled 7.91%. By Thursday morning the split had crossed the Pacific: Samsung Electronics fell 7.5% and SK Hynix dropped 9.5% in Seoul, while Nasdaq futures slipped 0.7% as Dow futures rose 0.4%. Investors were pricing two different futures at once: a broad market relieved by political gridlock and cheaper oil, and a chip sector suddenly asked to defend the pace of the AI buildout against a new, skeptical Congress.
The Rally and the Exception
Democrats won control of both the House and the Senate in the November 3 elections, leaving President Trump with a Republican White House facing a Democratic Congress when the new terms begin in January 2027. Markets rallied on the removal of election uncertainty and on the prospect of policy gridlock, the condition in which sweeping tax, spending, and regulatory changes become hardest to pass. The S&P 500 has risen after each of the last 11 midterm elections between 1982 and 2022, and one year out the index has returned 13.7% on average.
But the rally carried a price, and it ran through energy and interest rates. Investors raised their bets that a Democratic Congress would increase pressure on the Trump administration to end the war with Iran, and oil tumbled in response. West Texas Intermediate crude settled at $88.12 a barrel, down $7.97, or 8.29%, while Brent crude fell 8.67% to $94.62, both hitting two-month lows. The 10-year Treasury yield dropped 6.4 basis points to 4.886%. Falling yields and cheaper energy lifted rate-sensitive and consumer-facing parts of the market. The Dow, weighted toward industrials and financials, outperformed the tech-heavy Nasdaq-100, which rose just 0.91%. The energy sector itself was not spared: the S&P 500 energy group fell 4.8% as crude collapsed.
Then there were the chips. Semiconductor stocks had already been under pressure before the vote. Since October 6, chip and memory stocks have erased roughly $3.7 trillion in market value, with Intel down about 47% and Nvidia down about 16% over that stretch. The Philadelphia Stock Exchange Semiconductor index has fallen about 20% from its late-June peak. The election did not start the chip selloff; it redirected it, attaching a political question to a valuation problem that was already in motion.
Why the Rally Skipped the Chip Aisle
The market's logic was straightforward, and it ran through power and permitting. Data centers are both power-hungry and capital-intensive. If oil falls because the market expects the war, and the wartime urgency behind data-center approvals and grid interconnections, to wind down, then the expected speed of AI infrastructure deployment slows with it. The companies that sell the picks and shovels for that buildout trade on the slope of future demand, not the stock of installed capacity. A flatter slope is a valuation problem even when end demand is still growing.
The political channel reinforces the economic one. Strategists expect Democratic victories in Congress and in state governor contests, including in Texas, to hold back how quickly new data centers get built. Wells Fargo estimates Texas holds 27% of planned U.S. data-center gigawatts. A data-center slowdown should hit estimates for AI infrastructure stocks as well as cloud companies, UBS strategists said in a research note. The Texas detail matters even with a caveat: Republicans retain the majority in the state legislature, which limits how much a Democratic governor can change state policy on her own. The point is not that construction stops. It is that the margin of safety around the buildout's timing has narrowed.
That is why Microsoft and Amazon, the hyperscalers that pay for the data centers, could hold up while Nvidia, Micron, and Marvell fell. The buyers of compute benefit from cheaper power and a less frantic buildout. The sellers of compute equipment are judged on how much capacity gets installed and how fast.
The AMD Exception, and What It Proves
Not every chip name fell. AMD rose 1.01% on November 4, a day after reporting a record quarter that served as a company-specific counterweight to the wider concerns. Revenue reached $13.4 billion, up 45% year over year, with data-center revenue nearly doubling to $8.5 billion, up 97%, driven by EPYC processors and the launch of the Instinct MI450 GPUs. Non-GAAP earnings per share rose 74% to $2.09, and the company guided fourth-quarter revenue to approximately $17 billion, implying 66% annual growth at the midpoint.
"We delivered a record quarter with revenue exceeding the high end of our guidance as Data Center segment sales nearly doubled year-over-year," AMD chair and CEO Dr. Lisa Su said. "Demand for our Instinct and EPYC products continued to strengthen, and we expect growth to accelerate further in the fourth quarter as we scale Helios and 'Venice' production."
The divergence matters because it tells you what the market is doing. It is not rejecting AI demand itself; it is repricing the risk around the pace of deployment and the concentration of that demand in a handful of hyperscalers whose budgets could face new oversight. AMD's data-center revenue is up 97% because the customers are still spending. The selloff in the broader chain is a bet on what happens if those customers slow down.
Cyclical or Structural? The Call That Decides the Trade
This is where the analysis has to take a side, because the two answers point to opposite trades.
The cyclical case is strong on the surface. Semiconductor cycles are inventory and capex cycles, and history says post-midterm periods are kind to chips. Chips have climbed in all eight midterm cycles since 1994, measured from the last October close through the following June. In 2002 and 2018 they first fell another 11% before recovering, a reminder that the pattern does not guarantee a clean start. The AI demand story is not broken: AMD's numbers prove hyperscaler capex is still expanding, and enterprise AI adoption is in its early innings. If this is a cyclical leg, the right move is to buy the dip, and the 20% drawdown in the semiconductor index is the kind of air pocket that mean-reverts.
The structural case is harder to dismiss than it was six months ago, because three things have changed at once. First, the financing question: the buildout has been funded by a mix of hyperscaler balance sheets, private credit, and equity issuance, and rising scrutiny of how AI infrastructure is paid for raises the cost of capital for the marginal project. Second, the permitting question: a Democratic Congress with oversight powers can slow data-center approvals, environmental reviews, and power-grid interconnections without passing a single law. Third, the return question: after $3.7 trillion in value has been erased, the market is no longer willing to underwrite growth at any price. The bar for new capacity has moved from "will there be demand?" to "will the returns clear the cost of capital?"
The verdict is that this is a cyclical downturn layered on top of a structural repricing. The cycle, driven by inventory, capex timing, and post-election positioning, will revert. The valuation regime will not fully revert to what existed before the October rout, because the financing and permitting risks are now visible and priced. That distinction is the difference between a trade and an investment thesis.
The Second-Order Effect the Market Is Not Pricing
The first-order read is simple: chips fall, everything else rises. The second-order read is more interesting, and it runs through the bond market, the dollar, and the rest of the AI trade. If oil stays down and the war-risk premium drains out of energy prices, inflation expectations fall. That pulls Treasury yields lower, which supports long-duration growth stocks, the very names that led the AI rally. But there is a catch: lower yields also make the chip sector's high multiples look less painful, which should cushion the selloff. The fact that chips kept falling while yields dropped tells you the market is not worried about discount rates. It is worried about cash flows.
The cross-asset transmission runs further. A Democratic Congress means more oversight of financial institutions, emerging markets, and corporate relationships with the Trump administration. That raises the risk premium on sectors that benefited most from deregulation, not just chips but parts of financials and energy. Meanwhile, the rotation inside the AI trade is visible in the winners: clean-energy and grid-infrastructure names benefit if the buildout continues under different rules, while crypto-friendly policies face congressional opposition. Bitcoin briefly dropped below $76,000 as the legislative path for a U.S. crypto market-structure bill grew more uncertain. The market is rotating within the AI trade, not abandoning it.
The Strongest Case Against This View
The counter-thesis is that the market is overreacting to a political outcome that changes less than traders think. A divided government is the condition America has lived with for much of the past four decades, and it has not historically prevented corporate investment. The AI buildout is being driven by private companies, not government permits, and the hyperscalers have multi-year capex commitments that will not be unwound by a change in congressional committee chairmen. AMD's 97% data-center growth and its $17 billion fourth-quarter guide are the evidence: demand is real, visible, and accelerating. History backs the optimists: the S&P 500 has risen after each of the last 11 midterm elections, and the post-election year favors risk assets broadly.
Jeff Hirsch, editor of the Stock Trader's Almanac, bought the selloff on October 30 on exactly this logic. "Historically, seasonally, it sets up great for this period of time, the best months of the year for semiconductors," he said. Adam Johnson, founder of the Bullseye Brief and portfolio manager of the American Ingenuity Fund, argued the selloff had already priced in the worst. "The market kitchen-sinked it," Johnson said. "They discounted everything that could go wrong. And when you do that, typically that's when you see a bottom."
The answer to the counter-thesis is that the financing structure of this buildout does not match any prior technology cycle. Previous booms were funded largely from operating cash flow and public equity. This one leans on private credit and project finance tied to power and permitting, which is why oversight and interconnection queues matter more than they did in 2018 or 2002. The falsifying signal is concrete: if hyperscaler capital-expenditure guidance for 2027 holds or rises at the next earnings round, and if data-center power-interconnection queues do not lengthen through the first quarter of 2027, then the structural-repricing thesis is wrong and this is purely cyclical. Watch the January earnings calls for capex guidance from Microsoft, Meta, Alphabet, and Amazon.
What Comes Next
The market has drawn a line between the buyers and the sellers of AI infrastructure, and that line will define the next leg. In the short term, sentiment and positioning dominate: the 20% drawdown in semiconductors leaves room for a technical bounce, especially if the post-midterm rally broadens. In the medium term, fundamentals decide: companies with visible demand, like AMD with its 97% data-center growth and $17 billion guide, will separate from names whose valuations depended on an uninterrupted buildout. In the long term, the structural question of how the AI buildout is financed and permitted will set the sector's valuation ceiling.
Three scenarios frame the path. The base case is that the chip selloff stabilizes as earnings confirm hyperscaler demand is intact, and the sector grinds higher into 2027 alongside the broader post-midterm rally, but at lower multiples than in 2025. The upside case is that a clean resolution of the Iran conflict keeps oil low, yields fall further, and capex guidance comes in strong, producing a relief rally that takes semiconductors back toward their peaks. The downside case is that congressional oversight slows permitting, hyperscalers trim capex, and the $3.7 trillion drawdown extends into a deeper, multi-quarter de-rating.
Global crosscurrents add noise to the setup. The Bank of England voted 8-1 on Thursday to raise its benchmark rate by a quarter point to 4%, its first increase since 2023, as the conflict in the Middle East pushed energy prices higher and inflation above the central bank's forecast. That is a reminder that even as oil falls on peace hopes, global central banks are still fighting energy-driven inflation, and their reaction functions have not turned dovish in unison.
The market is not pricing the end of AI. It is pricing the end of AI at any price.
Explore more exclusive insights at nextfin.ai.
