Wall Street opened the week’s last session in the red on October 30, 2026, and the damage landed squarely on AI spending. According to Reuters, the Nasdaq Composite was down more than 2% at 9:42 a.m. ET after Amazon and Intel gave investors fresh reasons to question how fast the AI buildout pays back. Reports from Barron’s and Yahoo Finance put the index at its lowest opening level in more than two months, which is the figure this article uses instead of the “three-month low” wording some headlines floated.
The two stories are different but they rhyme. Amazon held its 2026 capital-expenditure forecast at roughly $220 billion while missing on third-quarter revenue and earnings. Intel warned it would burn cash in 2026 and watched its stock fall between 25% and 29%. We tracked the early-morning futures slide in our earlier report on Nasdaq futures and the $950 billion exposure, and this piece looks at what the cash-flow numbers mean for the companies that actually build and sell AI models.
Model labs do not appear on the Nasdaq tape this morning, yet they sit at the end of every dollar Amazon and its peers spend. Training runs, inference clusters and agent products all rent or buy the capacity that these capex plans create. When investors flinch at the bill, the pressure travels downstream to pricing, release cadence and who gets funded. For wider context on the sector, see our AI and machine learning coverage.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Happened to the Nasdaq on October 30
The morning tape was lopsided. Reuters reported the Nasdaq Composite down more than 2% by 9:42 a.m. ET, and other outlets described an open at the lowest level in more than two months. Amazon shares fell about 3% to 4% in reported Friday trading, while Intel dropped between 25% and 29% after its results and outlook. The spread in the Intel range reflects different snapshots from different outlets, so treat the exact close as unsettled until the session ends.
Two companies with very different businesses drove the same worry. Amazon is the largest cloud spender in the group and the home of AWS. Intel is a chipmaker trying to rebuild manufacturing capacity while the PC market softens. Investors read both reports as evidence that AI-era infrastructure costs arrive first and cash returns arrive later, if at all on the timeline the market had priced in.
Note what the reports do not say. No outlet in the supplied coverage quoted a named executive explaining the reaction today, so any claim about management intent here comes from filings and earnings summaries, not from fresh remarks. We flag that gap again in the quotes section below, where earlier public statements stand in as context rather than as reaction.
Amazon’s $220 Billion Capex Forecast Under Scrutiny
Amazon kept its 2026 capital-expenditure forecast at approximately $220 billion, according to Reuters and other reports. It also reported a third-quarter miss on both revenue and earnings. That combination is what rattled the room. A company can defend heavy spending when growth is surprising to the upside. It has a harder job when the quarter disappoints and the spending plan stays put.
The scale is easy to underrate. Amazon spent approximately $77 billion on property and equipment from July through September, against $35 billion in the same quarter a year earlier. That is more than double in twelve months, and it covers a single quarter. Reuters also reported a $94.8 billion year-over-year increase in net purchases of property and equipment on a trailing basis.
Quarterly spend: $77 billion versus $35 billion
Doing the arithmetic on the reported figures, Amazon added about $42 billion of quarterly property and equipment spending year over year, a rise of roughly 2.2 times. For a cloud business, that spending buys servers, accelerators, networking gear and the buildings and power that house them. AWS has to fill that capacity with paying workloads, and a large share of those workloads come from model developers and the enterprises building on their APIs.
Amazon’s own business case rests on contracted demand. Our look at Anthropic’s cloud marketplace revenue across three clouds shows how much of a model lab’s sales now flow through hyperscaler storefronts, which ties the lab’s growth to the cloud’s capacity plans.
The $42.4 Billion Free Cash Flow Hole
Yahoo Finance and Investing.com both highlighted the number that investors kept returning to: Amazon’s trailing 12-month free cash flow stood at negative $42.4 billion. Free cash flow is operating cash flow minus capital spending, so a negative figure of that size says capex is outrunning the cash the business throws off. We covered the metric in depth in Amazon’s negative $42.4 billion free cash flow, and the new session adds a market verdict to the arithmetic.
Negative free cash flow is not a crisis on its own. Growth companies run it for years. The problem is the combination of a missed quarter, an unchanged forecast and a price tag that keeps rising. Funding the gap means debt, equity or cuts elsewhere. Each of those has a cost that shows up in the share price, and today’s decline suggests holders are starting to price it in.
For model developers the key question is who absorbs the gap. If hyperscalers must defend margins, they may push harder on pricing for reserved capacity, tighten commitments, or favor customers who sign longer contracts. Labs with strong revenue can live with that. Labs still burning cash to train the next generation feel it first.
Intel’s 25% to 29% Drop and the Cash-Burn Warning
Intel’s move was the bigger single-stock shock. Shares fell between 25% and 29% after the company reported results and an outlook that included a warning it would burn cash in 2026. Reuters and Barron’s attributed the warning to three pressures: weak personal-computer demand, manufacturing bottlenecks and heavy spending to expand production. Our earlier note on Intel’s $15 billion to $16 billion fourth-quarter outlook covers the guidance details.
Why a PC-chip story belongs in an AI story
On its face, a weak PC market is not an AI problem. The link is the buildout itself. Intel is spending to expand production at the same time that its legacy demand softens, so it is trying to fund new capacity from a shrinking base. That mirrors the larger pattern. Capacity gets built on the assumption that AI demand will fill it, and the existing business has to carry the bridge.
Supply-chain tightness adds a second thread. When manufacturing bottlenecks hit one vendor, the rest of the stack feels it in lead times and prices, a theme we followed in the report on TSMC’s weighed 20% CoWoS price hike for Nvidia chips. Model labs buy accelerators at the end of that chain, so each delay or price bump raises the cost of every training run.
Why Model Labs Sit Behind the Spending Numbers
Capex headlines rarely name the end customer, but the demand story is plain. Frontier model developers need clusters for pretraining, reinforcement learning and serving. Agent products add steady inference load that never switches off. Every hyperscaler plan, from Amazon’s $220 billion to its rivals’ multibillion-dollar budgets, is a bet that this demand keeps compounding.
That is why the model race and the capex race now move together. Labs compete on release cadence, and cadence depends on compute. Readers tracking that contest can compare the newest releases in our breakdown of Gemini 4 Carbon against Claude Opus 5.5 and GPT-6.1 Terra. The supplied reporting for today’s selloff names no new model release as a trigger, so we make no claim that a specific launch drove the move.
The link also runs through ownership. Hyperscalers hold stakes in labs and book part of the lab’s spending as their own revenue. Our report on Anthropic’s Big Tech backers lays out how closely Amazon and Alphabet are tied to one model developer, which makes a spending scare at Amazon a relevant event for that lab’s financing outlook.
The Numbers at a Glance
The table below collects the figures from today’s reporting. Each row carries the outlet group that supplied it, and the final column states how firm the number is.
| Metric | Reported figure | Outlets | Status |
|---|---|---|---|
| Nasdaq Composite, 9:42 a.m. ET | Down more than 2% | Reuters | Intraday, will change |
| Nasdaq open | Lowest in more than two months | Barron’s, Yahoo Finance | “Three-month low” not confirmed |
| Amazon shares | Down about 3% to 4% | Reuters, The Economic Times | Intraday range |
| Intel shares | Down 25% to 29% | Reuters, Investing.com | Outlets differ |
| Amazon 2026 capex forecast | About $220 billion | Reuters, Barron’s | Maintained |
| Amazon trailing free cash flow | Negative $42.4 billion | Yahoo Finance | 12-month figure |
| Amazon net property and equipment purchases | Up $94.8 billion year over year | Reuters | Trailing basis |
| Amazon Q3 property and equipment spend | About $77 billion vs $35 billion | Investing.com | July to September |
What Big Tech CEOs Have Said About Overinvesting
None of the supplied coverage carries fresh on-record remarks from executives today. Earlier public statements still frame the argument, and we cite them as background, not as reaction. Amazon CEO Andy Jassy defended the company’s spending plan in remarks published by Amazon, saying: “We have customer commitments that make our capex investments predictable. We’re not investing approximately $200 billion in capex in 2026 on a hunch.” (Amazon). That framing leans on contracted demand, which is exactly what the market is now testing against a missed quarter.
Alphabet CEO Sundar Pichai has argued the opposite risk is larger. In remarks collected by Kiplinger, he said: “The one way I think about it is when you go through a curve like this, the risk of underinvesting is dramatically greater than the risk of overinvesting for us here.” (Kiplinger). He added that the infrastructure has “long useful lives” and can be applied across products, which is his answer to the stranded-asset worry.
Meta’s Mark Zuckerberg has been more candid about the downside. Kiplinger’s roundup records him saying: “If we end up misspending a couple of hundred billion dollars, I think that that is going to be very unfortunate, obviously.” (Kiplinger). Anthropic CEO Dario Amodei described the structural dilemma: “I think there’s a real dilemma deriving from uncertainty in how quickly the economic value is going to grow and the lag times on building the data centers that drives this.” (Kiplinger).
Read together, the four statements bracket the debate. Jassy points to contracts, Pichai to asymmetric risk, Zuckerberg to the cost of being wrong and Amodei to timing. Today’s market action says investors are leaning toward Amodei’s point: the lag between building and earning is the part that hurts.
Historical Context: Infrastructure Booms and Their Hangovers
Every large infrastructure cycle in tech has a version of this morning. The late-1990s telecom and fiber buildout laid far more capacity than demand needed in the short run, and investors punished the builders before the traffic caught up. The later cloud buildout went the other way, with spending that looked excessive eventually absorbed by workloads nobody had planned for. Which history applies to AI is the open question.
The difference this time is concentration. A handful of companies carry most of the spending, and their capex plans now sit at levels that dominate index-level earnings arguments. When Meta’s quarterly capital outlay surfaced earlier this week, we compared it with search growth in Meta’s $34.91 billion capex versus Google Search growth of 12%. Amazon’s numbers push the same comparison into a larger frame.
Another difference is financing. Market reports this week pointed to tech companies using bond markets to help fund AI infrastructure, though the supplied coverage names no specific issuers. Debt-funded buildouts survive a slow quarter less easily than cash-funded ones, because interest payments keep arriving whether or not the new capacity is full.
Competitive Comparison: Who Is Spending and Who Is Exposed
The second table maps today’s two stories against the wider field. It uses only qualitative labels where the fact sheet gives no figures, so no cell below states a number we cannot source.
| Company | Role in the AI stack | Today’s signal | Main exposure |
|---|---|---|---|
| Amazon | Cloud capacity, lab investor | Capex held near $220 billion, Q3 miss | Negative free cash flow, payback timing |
| Intel | Chip manufacturing | Cash-burn warning, 25% to 29% drop | PC demand, bottlenecks, expansion costs |
| Hyperscaler peers | Cloud and in-house models | No figures in the supplied reporting | Same capex-to-revenue question |
| Model labs | Customers of the capacity | No direct price move today | Compute pricing, funding appetite |
| Accelerator suppliers | Chips and packaging | Watched for order commentary | Order pace if budgets tighten |
The reading is simple. Spenders and builders take the first hit. Labs are customers in this picture, so they feel the effect through pricing and funding terms, not through a ticker. For a fuller look at the supply side, our report on Nvidia’s $3 billion investment in Meta shows how the chip vendors also fund their own customers.
Market Impact on Chips, Cloud and Software
Index math drives much of the pain. Our earlier note said the Nasdaq-100 faced a market-value hit of roughly $950 billion in futures trading, and the cash session has now opened lower. Concentration works both ways. The same mega-caps that lifted the index for two years now pull it down together when the spending story wobbles.
Chip names face a particular double bind. Strong AI demand supports orders, yet any sign that buyers will slow spending hits the group fast. Cloud names face the margin question, since depreciation on a $77 billion quarter flows into income statements for years. Software firms that resell model access face the cost side, because inference bills scale with usage.
For a longer-run view of how the largest buyers are weighing growth against spending, see our analysis of Big Tech’s $750 billion AI bet and slowing core growth. Today’s tape fits that thesis: when core growth cools, investors look harder at capex.
What This Means for AI Model Pricing and Releases
Pricing is the likeliest channel. Model developers pay for compute and charge for tokens. If cloud providers face pressure to protect margins, the first lever is the price of reserved and on-demand capacity. Labs can pass some of that to customers, absorb some of it, or shift work to cheaper architectures. We cover the economics from the buyer side in our guide to what renting an Nvidia H100 costs per hour.
Release cadence is the second channel. Frontier training runs are among the costliest line items in the industry, and investors who start doubting returns ask labs to show revenue before the next big run. That does not stop development. It shifts emphasis toward efficiency, smaller models and products that earn money sooner.
Open-weight and Chinese-lab models also gain a talking point. Cheaper-to-run systems look more attractive to buyers when the cost of premium compute climbs. Nothing in today’s supplied reporting confirms such a shift, so we treat it as a scenario to watch, not as a finding.
What to Watch Next
Four indicators will show whether this is a one-day flush or a regime change. First, the Friday close for both stocks, since intraday ranges of 3% to 4% for Amazon and 25% to 29% for Intel will settle into single numbers. Second, any shift in Amazon’s capex language at conference appearances, particularly around next year’s spending. Third, credit markets, where wider spreads on tech bonds would confirm that financing costs are biting. Fourth, order commentary from accelerator and memory suppliers.
Also watch the labs. Funding announcements, compute contracts and price changes on model APIs will arrive over the coming weeks. A cluster of delayed deals or price cuts would suggest the pressure is moving downstream, while a string of large new commitments would argue demand still outruns supply.
Five Predictions for the Next 90 Days
These are our analysis, not reported facts, and they carry the usual uncertainty.
- Capex plans hold, with tighter wording. Amazon and peers keep headline budgets near current levels but stress contracted demand and return timelines in their language.
- More bond issuance, with more scrutiny. Tech borrowers keep using debt markets, and investors demand clearer disclosure on what each dollar buys.
- Compute pricing firms up for spot capacity. Providers favor long-term commitments over on-demand rentals as they defend margins.
- Labs lean harder on efficiency. Expect more emphasis on smaller, cheaper models and revenue-generating agent products over raw scale announcements.
- Intel’s story stays tied to execution. Manufacturing milestones, not AI narratives, set the stock’s direction until the cash-burn picture improves.
Bottom Line for Readers Who Build With AI
If you ship products on hosted models, today’s news does not change your API bill by lunchtime. It does raise the odds that cost, not capability, becomes the main selection criterion over the next two quarters. Build with provider flexibility, track token spend per feature and keep a fallback model in your stack.
If you invest, remember what the fact sheet confirms and what it leaves open. The Nasdaq fell more than 2%, Amazon kept a roughly $220 billion forecast, free cash flow stands at negative $42.4 billion and Intel warned of a cash burn. No reported fact tells us whether the buildout earns its keep. That answer arrives in cloud revenue lines over the next several quarters.
Frequently Asked Questions
Did the Nasdaq hit a three-month low on October 30, 2026?
Not according to the reports we can confirm. Barron’s and Yahoo Finance described an open at the lowest level in more than two months, and Reuters had the Composite down more than 2% at 9:42 a.m. ET. The “three-month low” phrasing in some headlines is unconfirmed.
How much is Amazon planning to spend on capital expenditure in 2026?
Amazon maintained a 2026 capex forecast of approximately $220 billion, per Reuters and Barron’s. Spending on property and equipment ran about $77 billion from July through September, compared with $35 billion a year earlier.
What is Amazon’s free cash flow?
Yahoo Finance reported negative $42.4 billion over the preceding 12 months. Free cash flow subtracts capital spending from operating cash flow, so the figure shows spending outrunning cash generation.
Why did Intel stock fall 25% to 29%?
Intel warned it would burn cash in 2026. Reuters and other outlets tied the warning to weak PC demand, manufacturing bottlenecks and heavy spending to expand production. Outlets reported different percentage drops, which is why we give a range.
Why does an AI capex scare matter to model developers?
Model developers rent or buy the capacity that capex creates. If providers face pressure to protect margins, compute pricing and funding terms can tighten, which affects training budgets and API prices. Today’s supplied reporting shows no direct price change at any model lab.
Did a new AI model release cause the selloff?
No supplied report links today’s move to a model launch. The cited drivers are Amazon’s missed quarter and unchanged capex plan and Intel’s cash-burn warning.
Should developers change their AI stack because of this news?
Not immediately. A sensible step is to avoid single-provider lock-in and track cost per request, since the cost of capacity is the part of the story most likely to reach your bill.
