The market entered election week on Monday.
Volatility is nothing unusual at this time of year, but this time there are more factors in play. There is only one major piece of macroeconomic data: the October jobs report, due out on November 6. There are also few major earnings reports, with CoreWeave and companies related to “Gemini” among the two worth watching. But more unsettling than macro data and earnings are the midterm elections themselves—and whether Iran will make a move during this window.
The U.S. midterm elections are scheduled for November 3. All 435 seats in the House of Representatives, 35 Senate seats, and the governorships of 36 states are up for election. Polls suggest Democrats are likely to win the House, while control of the Senate is up for grabs. Xinhua reports that the “tariff war” launched by the Trump administration and the conflict with Iran have significantly increased the cost of living for Americans. Foreign policy has unusually become an important factor shaping this year’s election. In other words, oil prices are no longer just a variable for energy traders—they are being priced in at the ballot box.
Iran’s “Two-Day Plan” and a Weekend Signal That Failed
Iran has again floated a plan to “open the Strait of Hormuz for two days,” but the U.S. response was cool. The window for talks is indeed narrowing. In an analysis published on November 2, CCTV News noted that Iran treats the right of passage through Hormuz as a tradable asset, while the U.S. believes international waterways should be open in the first place and does not accept that this is something to trade away. The two sides are diametrically opposed on the question of whether sanctions should be eased first or nuclear issues discussed first, making direct talks in the near term unlikely.
More concerning is the timing. Reports say the U.S. has informed its Gulf allies that it could launch a new round of strikes against Iran as early as November 4—the day after the midterm elections. Trump himself has once again named the underground nuclear facility at Pickaxe Mountain as a possible target for a limited strike. The likelihood of an all-out war is low, but several analysts see a symbolic precision strike within days or weeks of the election results becoming clear as the base case.
Oil prices have responded honestly to this signal. Intraday on November 2, prices briefly fell by more than 2%, but Brent remains anchored in the $110–113 range, and the market has not factored the lifting of the blockade into its base case. The weekend futures curve shows no sign of improvement. If Iran turns up the pressure a little during election week and pushes oil prices higher, Republicans will be in a very difficult position—voters are already casting their ballots based on the cost of living, and the Trump administration will struggle to find a clean scapegoat on this issue.
AI’s Credit Fault Line: The Shift from Narrative to Cash Flow
U.S. stocks had already set an uneasy tone for election week last Friday. The Nasdaq Composite fell 2.86%, the Philadelphia Semiconductor Index dropped 7.2%, and Intel plunged 28.58% in a single session, its biggest one-day drop in more than four decades. The immediate trigger was Intel’s far weaker-than-expected fourth-quarter revenue guidance, compounded by soft PC demand, but the deeper signal lies in credit markets.
CoreWeave offers a window into the structure of AI financing. As of the second quarter of 2026, it had about $35 billion in interest-bearing debt, 29% of which was unsecured credit debt, and 90% of its debt had recourse to the parent company. Yields on its unsecured bonds have risen to around 15%, while five-year CDS spreads have topped 1,100 basis points, implying an approximately 60% cumulative probability of default over five years. CICC describes this type of Neocloud issuer as the “subprime” weak-credit link in the AI financing system: net debt/EBITDA is about 7.5 times, compared with just 0.1 times for a sample of large tech companies. The cost of secured DDTLs may rise less than that of unsecured debt, but if the parent company’s credit deteriorates, financing terms for new project loans will tighten as well.
The model companies are also under pressure. Anthropic’s quarter-over-quarter revenue growth slowed from about 145% to 45% in the third quarter, and some shareholders and prospective IPO investors are discussing lowering its expected listing valuation from around $2 trillion to $1.5 trillion. Blackstone has revised its expectations, believing Anthropic will most likely remain in the BB category after its IPO—the upper end of junk grade—rather than achieve the investment-grade rating previously hoped for. KBRA gave it a BB+ rating in June, just one notch below the lowest investment-grade rating of BBB-, but that single notch marks the boundary for whether pension and insurance funds can hold the debt at scale. As of the end of June, Anthropic held more than $130 billion in cash and short-term investments, but computing and infrastructure operating expenses reached $8.945 billion in the first half, or 55% of revenue. Minimum purchase commitments totaled about $269.4 billion, with payment obligations extending beyond 2033. The money is still there, but the pace and rigidity of its spending are being repriced by credit markets.
This is not a story about the “end of the AI narrative.” Credit at large tech companies remains solid: Microsoft generated $18.6 billion in free cash flow even as its capital expenditure reached $45.2 billion for the quarter. But a divide has emerged. Companies with current cash flow and access to low-cost financing are being placed in one drawer, while those relying on long-dated contracts and highly leveraged expansion are being placed in another. Intel’s plunge, CoreWeave’s spreads, and expectations for Anthropic’s rating are all sounds made as that same drawer is pulled open.
The Market Is Rotating, Not Exiting
Last week’s losses could easily be misread as a wholesale retreat from risk. But break down the indices and the picture looks different: the Nasdaq fell 2% for the week, the S&P 500 lost 0.57%, while small caps in the Russell 2000 actually gained 1.3%. Money has not left the market. It has moved away from the most crowded trades, those most dependent on future cash flows, and toward current earnings and lower valuations.
Next week’s test will not be the election news itself, but whether several price signals change together in its aftermath. The first is market breadth: if the Russell 2000 and the equal-weighted S&P continue to outperform the Nasdaq, the rotation is still taking place within the market. If small caps, financials, and consumer stocks also start weakening in tandem, the localized repricing of tech stocks could spread into a broader contraction. The second is whether AI earnings can translate “orders” into “margins and cash flow.” Apple, Palantir, AMD, Arista, and Qualcomm will report around the election, covering four parts of the chain: AI software, data-center chips, cloud networking, and consumer electronics. The previous earnings season already showed that the market no longer rewards revenue growth alone; “growth” without support from margins, cash flow, and next-quarter guidance will be discounted quickly. The third is the jobs report. The U.S. added just 29,000 jobs in September, with unemployment at 4.2%. In October, the Fed held rates at 3.75%–4.00%, with two policymakers favoring a 25-basis-point hike. A combination of moderate employment growth and contained wages is most favorable for tech valuations; if employment and wages are both stronger, long-term yields will put further pressure on high-valuation assets; if employment weakens significantly, rate pressure will ease, but earnings expectations will also have to be revised down.
Only improvement in all three would count as stabilization. Meeting just one of the conditions would most likely amount to no more than a rebound.
A Bitcoin Bull’s Weekend, and the Frustration of Working with AI
When I put a Bitcoin order in around $76,000 on Friday, I still thought there might be an opportunity. I’m less confident now—not because my view has changed, but because there are more constraints around it. Binance withdrawals have begun, and sentiment among crypto investors is indeed easing. But for BTC, oil prices and tech stocks are two external variables that cannot be ignored. As long as both hold steady, the price action should be acceptable. The problem is that oil prices are giving me a headache right now.
I’m still shorting, but funding rates are too high. I trimmed my position a little during the recent choppy period, but my core view hasn’t changed: the probability of a substantive reconciliation before the midterm elections is very low. There is more than one technical exchange of concessions separating Iran’s demands from the U.S. bottom line.
A more specific sense of exhaustion comes from something else. The monitoring project itself is running well, but a bug in the sandbox’s write permissions trapped two AIs at the same time. ChatGPT simply couldn’t get past it, couldn’t find a workaround, and even automatically reported the bug to OpenAI before just stopping there. The other model was worse: despite being completely unsure, it sent ChatGPT looking for a menu that didn’t exist. When called out, it started apologizing, then began making things difficult for ChatGPT. The computer ran all night as the two models burned through tokens in a loop without getting a single thing done. In the end, the next morning I manually asked ChatGPT to start a new Work session to test write permissions. The test passed, and only then could we continue.
There is a curious similarity between this and the market during election week: an individual model—or an individual signal—may appear to be “working,” but the real bottleneck is often at a level it cannot see. Oil prices affect tech stocks; credit spreads affect AI capital spending; election results affect data-center permits. Texas has already frozen TCEQ environmental permits, while most of the 62.2 GW of net new capacity planned for 2028 has yet to break ground and is concentrated in states holding gubernatorial elections on November 3. Construction for 2027 is essentially locked in; elections cannot change concrete that is already being poured. But the 2028 pipeline is still moving through permitting and power infrastructure processes, and the shape of the rules will determine which projects survive.
When two AIs have trouble working together, at least you can see where each one is stuck. The market offers no such convenience. All you can do amid the noise of election week is distinguish the facts being priced in from the narratives being repeated. Nonfarm payrolls, credit spreads, market breadth, the level around which oil prices are anchored—none of these need a vote, but they will tell you where the money actually went after the voting is over.
This article is for sharing information and views only and does not constitute investment advice.