AI Sector Correction:
What the Market Is Repricing
Stocks have corrected,
but AI demand keeps growing
From October 29 to November 6, the U.S. tech sector went through two waves of selling. The Nasdaq entered a correction, and the semiconductor index lost 12% on November 5 alone. Quarterly earnings, meanwhile, showed cloud revenue accelerating.
Below, we look at what exactly the market is repricing and how it affects public and private AI companies.
Key Takeaways
AI demand is growing at a record pace
Revenue grew 103% at Google Cloud, 48% at Azure (Microsoft) and 39% at AWS (Amazon). The three largest cloud providers have a combined contract backlog of more than $1.8T.
AI economics and the pace of the infrastructure build-out are weighing on stocks
Demand remains intact. Meanwhile, legacy business models are losing ground, investment is outpacing cash flow and, since the election, data center construction faces a new risk.
Analyst sentiment remains positive
Most analysts rate Alphabet, Amazon, Meta and ServiceNow a “buy,” and average price targets imply 35–48% upside.
Control of Congress is shifting to the Democrats
After rallying on November 4, the market reversed: on November 5, the semiconductor index fell 12% on fears that data center construction will slow. We do not expect a direct impact on the IPO market.
The main beneficiaries of AI remain private
Their revenue is expanding several-fold and they are reaching profitability, but stock indexes do not reflect this growth.
What Happened in the Market
The sell-off began on October 29 with Alphabet’s earnings and deepened on October 30 after Intel and Amazon reported. Companies in the Nasdaq-100, the 100 largest nonfinancial stocks listed on the Nasdaq, lost more than $2T in market value in October. A second, stronger wave of selling followed on November 5, after the election.
Indexes · October 30
Biggest Moves
Share price declines, October 29–30
| Name | Value | Note |
|---|---|---|
| Intel | −28.6% | on October 30, still +78% year to date |
| ServiceNow | −27% | on October 29, its worst day on record, and −31% for the week |
| Alphabet | −12.6% | on October 29, its worst day on record |
| Amazon | −5.6% | over two sessions |
| Nvidia | −4.2% | on October 30 |
| Meta | −1.6% | on October 30 |
| Microsoft | −1.6% | −0.2% on October 29 and −1.6% on October 30 |
The Sell-Off in Context: Gains Since 2024
Share price, dashed line 100 = December 29, 2023 close
| Date | Nvidia | Alphabet | Amazon | Intel |
|---|---|---|---|---|
| Dec 2023 | 100 | 100 | 100 | 100 |
| Jan 2024 | 124 | 100 | 102 | 86 |
| Feb 2024 | 160 | 99 | 116 | 86 |
| Mar 2024 | 183 | 108 | 119 | 88 |
| Apr 2024 | 174 | 117 | 115 | 61 |
| May 2024 | 221 | 123 | 116 | 61 |
| Jun 2024 | 249 | 130 | 127 | 62 |
| Jul 2024 | 236 | 123 | 123 | 61 |
| Aug 2024 | 241 | 117 | 117 | 44 |
| Sep 2024 | 245 | 119 | 123 | 47 |
| Oct 2024 | 268 | 123 | 123 | 43 |
| Nov 2024 | 279 | 121 | 137 | 48 |
| Dec 2024 | 271 | 135 | 144 | 40 |
| Jan 2025 | 242 | 146 | 156 | 39 |
| Feb 2025 | 252 | 122 | 140 | 47 |
| Mar 2025 | 219 | 111 | 125 | 45 |
| Apr 2025 | 220 | 114 | 121 | 40 |
| May 2025 | 273 | 123 | 135 | 39 |
| Jun 2025 | 319 | 126 | 144 | 45 |
| Jul 2025 | 359 | 137 | 154 | 39 |
| Aug 2025 | 352 | 152 | 151 | 48 |
| Sep 2025 | 377 | 174 | 145 | 67 |
| Oct 2025 | 409 | 201 | 161 | 80 |
| Nov 2025 | 357 | 229 | 153 | 81 |
| Dec 2025 | 377 | 224 | 152 | 73 |
| Jan 2026 | 386 | 242 | 158 | 92 |
| Feb 2026 | 358 | 223 | 138 | 91 |
| Mar 2026 | 352 | 206 | 137 | 88 |
| Apr 2026 | 403 | 276 | 174 | 188 |
| May 2026 | 426 | 272 | 178 | 228 |
| Jun 2026 | 404 | 256 | 157 | 278 |
| Jul 2026 | 405 | 255 | 179 | 180 |
| Aug 2026 | 446 | 243 | 171 | 178 |
| Sep 2026 | 461 | 246 | 164 | 239 |
| Oct 28–29 | 431 | 233 | 167 | 183 |
| Oct 30 | 413 | 200 | 158 | 131 |
| Nov 6 | 377 | 204 | 171 | 112 |
Monthly closes; the final points show the day before the sell-off (October 28 for Alphabet and Amazon, October 29 for Nvidia and Intel), October 30, 2026 and November 6, 2026.
Nvidia, Alphabet and Amazon have held on to most of their gains since 2024. Intel is the exception: its gain since early 2024 has shrunk to about 12%, though the stock is still up 53% this year. The indexes also remain above their early-2024 levels: the S&P 500 is up 55%, the Nasdaq 64%, and the semiconductor index has more than doubled.
Why Are Stocks Falling Despite Strong AI Demand?
The drivers differ from company to company, but none of the reports showed weaker demand for AI.
Alphabet and the New Economics of Search
Google Cloud revenue grew 103% to $30.7B. Growth in search revenue slowed to 12% from 17% in the second quarter, below the 15% the market expected.
Alphabet Quarterly Revenue by Segment
$B, Q4 2018 to Q3 2026
| Quarter | Search | Everything else | Google Cloud |
|---|---|---|---|
| Q4 2018 | 23.1 | 14.1 | 1.4 |
| Q1 2019 | 22.4 | 11.9 | 1.6 |
| Q2 2019 | 23.5 | 13.1 | 1.9 |
| Q3 2019 | 24.7 | 13.3 | 2.2 |
| Q4 2019 | 27.2 | 16.1 | 2.4 |
| Q1 2020 | 24.5 | 13.8 | 2.6 |
| Q2 2020 | 21.3 | 13.9 | 2.8 |
| Q3 2020 | 26.2 | 16.4 | 3.2 |
| Q4 2020 | 31.7 | 20.9 | 3.6 |
| Q1 2021 | 31.9 | 19.3 | 3.8 |
| Q2 2021 | 35.8 | 21.4 | 4.4 |
| Q3 2021 | 37.8 | 22.1 | 4.9 |
| Q4 2021 | 43.1 | 26.3 | 5.3 |
| Q1 2022 | 39.6 | 22.6 | 5.7 |
| Q2 2022 | 40.6 | 22.6 | 6.1 |
| Q3 2022 | 39.5 | 22.6 | 6.7 |
| Q4 2022 | 42.5 | 26 | 7.2 |
| Q1 2023 | 40.4 | 21.9 | 7.4 |
| Q2 2023 | 42.5 | 23.9 | 7.8 |
| Q3 2023 | 44.1 | 24.3 | 8.3 |
| Q4 2023 | 48 | 29 | 9 |
| Q1 2024 | 46.3 | 24.8 | 9.5 |
| Q2 2024 | 48.5 | 25.8 | 10.1 |
| Q3 2024 | 49.4 | 27.4 | 11.2 |
| Q4 2024 | 54 | 30.3 | 11.8 |
| Q1 2025 | 50.9 | 27.3 | 12.2 |
| Q2 2025 | 54.2 | 28.5 | 13.5 |
| Q3 2025 | 56.5 | 30 | 15.1 |
| Q4 2025 | 63 | 33 | 17.5 |
| Q1 2026 | 60.6 | 29.7 | 19.8 |
| Q2 2026 | 63.4 | 31.8 | 24.7 |
| Q3 2026 | 63.4 | 33.9 | 30.7 |
Search itself is not shrinking. Its revenue reached a record $63.4B, up 12% from a year earlier. Search’s share of total revenue fell to 49.7% from 55.3% because Google Cloud is growing many times faster. Compared with the second quarter, however, search revenue was virtually unchanged. Search queries are growing at the fastest pace in five years. However, AI now generates a growing share of the answers, and those answers do not yet earn as much advertising revenue as traditional search results.
The audience remains, while the monetization model is changing. For Google, this is a risk; for the AI cycle, it confirms a structural shift.
Microsoft, Amazon, Meta and the Payback Question
Revenue Is Accelerating
Microsoft’s contract backlog jumped 73% to $677B. Amazon expects AWS growth of more than 40% in both 2027 and 2028. At Meta, the owner of Facebook and Instagram, the average price per ad rose 14%.
Investment Is Growing Even Faster
Meta has $825B in future commitments, and Alphabet more than $1T. Cloud giants have issued $182B of bonds in 2026, compared with $17B in 2024, according to Bank of America. S&P lowered its outlook on Meta’s credit rating to negative.
Investor concerns center on the payback period for this spending. Demand itself is not in question.
Intel: Memory and Component Shortages
Quarterly results beat expectations. Revenue rose 21% to $16.5B, adjusted earnings per share (EPS) came in at $0.52 versus $0.40, and data center revenue jumped 78%, the strongest growth in 15 years. However, the fourth-quarter outlook fell short of the analyst consensus. There are two reasons.
Memory
Memory makers are shifting capacity to AI data centers. Intel CEO Lip-Bu Tan said customers are already signaling a slowdown in the PC market because of memory prices.
Server chip components
Demand exceeds production capacity, and the constraints could last until mid-2027.
The issue extends across the PC market. According to research firm IDC, PC shipments fell 20.1% in the third quarter to 62.7 million units, the steepest third-quarter decline on record.
Some demand was pulled forward into the first half through early purchases. According to Gartner, fewer affordable models are available, making it harder for mainstream buyers to upgrade their computers. Intel expects the PC market to shrink about 12% in 2026 and by more than 20% in the second half.
Apple was hit less thanks to long-term memory contracts. Mac shipments fell 11.3%, compared with drops of 30.9% at HP and 25% at Dell, and Mac revenue rose 24% to $10.8B. Even so, Apple has already raised prices and expects supply constraints for the Mac and iPad in the December quarter.
Profit-taking also played a role. Before the report, Intel shares had risen about 2.5-fold this year, so the weak outlook prompted investors to lock in gains.
Intel’s results reflect the challenges in the PC market, where the AI boom is absorbing memory production capacity. They are unrelated to AI demand.
ServiceNow and the Risk of AI Agents Replacing SaaS
Shares of ServiceNow, a platform for automating corporate workflows, fell 27% in a single day, the worst drop in the company’s history. The trigger was earnings and an outlook for contracted revenue that came in slightly below expectations. The market interpreted this as a sign that AI agents (software that carries out tasks on its own) are starting to displace SaaS, the business software sold by subscription. The bar remains high: according to Bloomberg, 90% of software companies in the S&P 500 beat earnings estimates, yet the market still reacted negatively.
Four forces are driving the correction: AI is displacing legacy business models, investment is outpacing cash flow, component shortages are constraining the scale-up and, since the election, politics poses an added risk to data center construction. None of them indicates the end of the AI cycle, and demand continues to grow.
Sources: IDC, Gartner, Intel and Apple earnings calls, Bank of America, Bloomberg.
The Analyst View
Most analysts kept their “buy” ratings on Alphabet, Amazon, Meta and ServiceNow. Average price targets imply 35–48% upside from the October 30 close. Intel is mostly rated “hold,” but its average price target of about $100 implies roughly 50% upside.
Expects cloud companies’ capex to keep rising in 2027–2028.
Expects the sector to improve in the first half of 2027.
Valuations Do Not Appear Stretched
The key question for the market is the payback period on this investment.
Source: FactSet.
Private Markets: AI Leaders Are Turning Profitable
Public giants are financing AI infrastructure, while many of its direct beneficiaries remain private.
Anthropic
On October 26, Anthropic, the developer of the Claude AI models, filed for an IPO (Form S-1) and later updated the filing with third-quarter results. The listing is expected in November. A date has not yet been announced.
Anthropic Revenue
Adjusted Operating Income
Based on our information, Anthropic’s current revenue run rate is about $85B. At the end of July, the company told investors it stood at $65B.
Adjusted income excludes some expenses. Under GAAP, the U.S. accounting standard, Anthropic posted a $567M operating loss in the third quarter because of expensive short-term computing capacity. Free cash flow for the first half was positive, at about $3.4B.
Other Private Leaders
Other private leaders are also posting strong growth.
Databricks
- Revenue run rate: more than $7B (Q2 2026)
- Growth: more than 80% YoY
- Positive adjusted free cash flow over the past 12 months
Higgsfield
- Revenue run rate: more than $1B (September 2026)
- 18 months since product launch
- Cash flow positive since May 2026
OpenEvidence
- Revenue run rate: about $300M (July 2026)
- Doubled in roughly seven months
- Gross margin: about 90%
- Cash flow breakeven
Crusoe
- 2026 revenue forecast: about $2B
- Valuation: $30.9B (September 2026)
- Profitability not disclosed
The companies shaping the AI market are mostly private. Public markets register the capex and the pressure on legacy business models, while the growth of the new leaders remains outside the indexes. Anthropic’s IPO could be the first major AI listing of the new cycle. Its timing, however, will depend on how quickly public markets stabilize.
Sources: Constellation Research, Higgsfield, Startup Fortune, PYMNTS, Sacra (OpenEvidence), Sacra (Crusoe), StockAnalysis.
Democrats Take Control of Congress
Democrats won control of both chambers of Congress in the November 3 elections, according to the Associated Press.
At least 54 seats to 44
The races in Mississippi and Alaska are still undecided. Even in the best case, Democrats would hold 56 seats, short of the 60 needed to overcome a filibuster, the procedural tactic the minority uses to block legislation.
237 seats, with 218 needed for a majority
Democrats have 237 seats, according to the AP. CNN projects at least 245.
On November 4, the market initially rallied: investors priced in legislative gridlock, under which abrupt policy changes are unlikely. The Dow Jones gained 3.03%, its best day after a midterm election since 1982, and the S&P 500 rose 1.58%.
The reaction within the AI sector was mixed. Large platforms gained 1–7%, while data center operators, related energy companies and chipmakers fell: Hut 8 by 23%, Micron by 7%. Nvidia lost 3.2% and ceded the title of the world’s most valuable company to Apple.
On November 5, the market reversed, and the sell-off became the steepest since April 2025.
Indexes · November 5
Companies · November 5
One-day share price declines
| Name | Value | Note |
|---|---|---|
| SanDisk | −22% | |
| AMD and Micron | ≈ −17% | about −17% each |
| IREN, Applied Digital, Nebius | > −14% | down more than 14% each |
| TSMC and Broadcom | −11 / −10% | −11% and −10% |
| Intel | −10.4% | |
| Nvidia | −9.26% | wiping out about $450B in market value |
| Meta, Microsoft, Amazon, Alphabet | −2…−4% | −2% to −4% |
The trigger was concern that the Democratic victory would slow data center construction. Meta President Dina Powell McCormick acknowledged that building them would become harder. Dan Ives of Yorkville Ives still expects Big Tech capex to top $1T in 2027, though he sees growth slowing in 2028.
Large platforms again fell less than the market: for them, slower construction means lower capex and more free cash flow. Bitcoin rose 3.5% on November 4 and lost almost 6% on November 5.
On November 6, the market recovered some of its losses. The semiconductor index rose 4.2%, AMD 11%, Intel 7.7% and Nvidia 2.8%. The Nasdaq gained 1.1% and the S&P 500 0.3%, while the Dow Jones slipped 0.5% as strong labor market data pushed bond yields higher.
For the week, the Nasdaq lost 3.4%, the S&P 500 2.7% and the Dow Jones 0.9%. It was the worst week for the S&P 500 and the Nasdaq since June.
Implications for the Tech Sector
Data centers are the main risk. Texas remains the key state: it is expected to host 25–30% of future capacity. According to AP projections, Democrats won both a Senate seat there (James Talarico) and the governorship (Gina Hinojosa).
Tighter rules in Texas were coming whatever the outcome. Talarico’s plan largely mirrors proposals from Republican Governor Greg Abbott. Talarico himself does not support a moratorium, while Hinojosa backed one over the summer, so her first decisions as governor will warrant close attention.
Local restrictions are already being adopted. A federal ban on construction would be difficult to enact, but according to a tally by the law firm Duane Morris, 38 of 39 local initiatives with known outcomes impose restrictions. For the industry, this primarily means higher costs and longer project timelines.
Calls the sell-off overdone and expects tighter rules mainly to push some projects from 2028 to 2029, stretching the AI investment cycle over time while keeping its size intact. This scenario assumes the restrictions stay local.
If moratoriums are imposed by several large states, which Goldman Sachs estimates account for about 45% of planned capacity, delays would be longer and the pressure on chipmakers and data center operators would persist.
We do not expect a direct impact on the IPO market. Listing rules are set by the Securities and Exchange Commission (SEC), which continues to make it easier to go public, so decisions by Congress will have little effect on them. The IPO market will still depend on interest rates, AI company valuations and how recent listings trade.
Volatility could have an indirect effect on IPOs. On November 5, the VIX, a gauge of expected S&P 500 volatility known as the fear index, rose to its highest level since March, and companies tend to postpone listings in unstable periods.
Government funding for technology may shift. Starting in January, Democrats will chair the appropriations committees in both chambers. Analysts expect them to take a more critical view of rising defense spending and to redirect part of technology funding toward their own priorities, above all clean energy, energy research such as ARPA-E, and science programs at the NSF and NASA.
The shift is likely to be gradual. Restoring repealed clean energy incentives without the president’s consent will be difficult. Support for critical materials production and for reducing dependence on China will remain bipartisan.
Macro factors are also still weighing on the market: PCE inflation, the Fed’s preferred gauge, at 3.6%; geopolitics; expensive energy; and expectations of a Fed rate hike. After strong labor market data on November 6, futures put the probability of a December hike at about 80%.
The Historical Record
Higher volatility
The S&P 500’s average maximum drawdown in midterm election years since 1960.
Recovery from the lows
The S&P 500’s average gain over the 12 months after a midterm-year low.
Gains after the election
The S&P 500 was higher 12 months after each midterm election.
Divided government is no obstacle
The S&P 500’s average gain over the year after elections that left no party in full control (14 of 15 cases).
The election uncertainty has been removed, but the market has begun to reprice AI infrastructure. The main risk now lies in the pace of data center construction, and it hits chipmakers and data center operators hardest. Big Tech capex is still expected to exceed $1T in 2027. The historical record is favorable, though it does not guarantee a repeat.
Vote counting continues in some districts. Sources: AP, CNN, Reuters, Al Jazeera, Fortune, Reuters (November 5), INDmoney (citing Duane Morris), Washington Technology, Akin Gump, Reuters (November 9), Yahoo Finance, CNBC, Seoul Economic Daily, The Block, Texas Tribune.
China’s AI Market: A Bubble or a Correction?
The STAR50 index has fallen more than 40% from its July peak. It tracks the 50 largest companies on the Shanghai Stock Exchange’s tech-focused STAR Market. Reuters characterized the decline as a burst AI bubble.
In early November, Beijing announced measures to support the market. The People’s Bank of China launched swap operations under which 31 brokers and asset managers requested more than 100 billion yuan ($15B) in liquidity, and regulators discussed with banks the terms for refinancing share buybacks.
We see a correction, not a burst bubble. There are four reasons.
1. A Few Companies Drove Much of the Decline
Cambricon, a developer of AI chips, accounted for nearly 10% of the index. Its market value topped 1 trillion yuan ($139B) in June, but third-quarter revenue came in almost 50% below expectations because of shortages of high-bandwidth memory (HBM) and restrictions on equipment. Privately held Huawei remains China’s main alternative to Nvidia. In 2025 it shipped 812,000 Ascend AI accelerators, about half of all shipments by Chinese makers and seven times as many as Cambricon.
2. The Largest Model Developers Aren’t Listed
Moonshot AI and DeepSeek are private. Within a few months, they reached annual recurring revenue (ARR) of about $1B, and their models rival those of leading Western labs. As in the U.S., most of the growth is not reflected in the indexes.
3. IPO Premiums Persist
vs. IPO Price
Even after the correction, the largest AI listings of 2025–2026 are trading well above their IPO prices.
| Name | Value | Note |
|---|---|---|
| Unitree | +346% | robots · 150.80 → 672.00 yuan (Oct. 26) |
| MetaX | +330% | AI chips · 104.66 → 450.00 yuan (Oct. 30) |
| Moore Threads | +182% | AI chips · 114.28 → 322.65 yuan (Oct. 30) |
IPO price → closing price. Sources: Investing.com, Channel NewsAsia.
4. More Mature Companies Are Going Public
Many AI companies, such as MetaX and Moore Threads, went public with annual revenue of about $100M or less and at valuations of 50–125 times revenue. China’s securities regulator, the CSRC, has already tightened IPO requirements for robotics companies. Moonshot and DeepSeek could list with ARR of $1–2B or more, an order of magnitude higher. Some analysts believe the wait for these listings is already drawing investors away from today’s publicly traded AI companies.
The correction in Chinese AI indexes does not indicate that the industry is slowing. The leaders are private, and, as in the U.S., value is created before the IPO.
A detailed analysis of the Chinese market is available in our separate report, which we can provide on request. On Beijing’s measures: Yahoo Finance.
Sources: Reuters, Investing.com, Channel NewsAsia.
The AI Cycle Continues as Market Structure Changes
The correction indicates neither the end of the AI cycle nor a burst bubble: demand is growing at a record pace. But the market no longer rewards all growth equally. Companies whose value depends on uninterrupted construction, namely chipmakers, memory makers and data center operators, are under pressure, while large platforms with real revenue and cash flow are falling far less. What is changing is the structure of the market.
AI is displacing the business models of previous cycles
Infrastructure investment is outpacing cash flow
Component shortages are limiting the scale-up
Investors are paying closer attention to payback periods
Volatility may persist, with several significant events ahead.
October inflation data and CoreWeave’s earnings
Nvidia’s quarterly earnings; on November 18, Jensen Huang has been asked to give Congress a closed-door briefing
Current U.S. government funding expires
The Fed’s rate decision
The main takeaway of the week is the widening gap between public and private markets. Many AI beneficiaries, including Anthropic, Databricks, Moonshot AI, DeepSeek and Huawei, remain private. Their revenue is expanding several-fold and some are already generating positive cash flow, but the indexes do not reflect this. Private markets are not insulated, though: private valuations follow their public peers with a lag, so the timing of exits in the coming months will depend on the market stabilizing. Anthropic’s IPO could become a catalyst for a new cycle and open the way to public markets for other fast-growing companies.
Many of the companies shaping new technology trends are private, and a significant share of their value is created before the IPO.
This material reflects the authors’ view as of November 9, 2026. Figures are based on company filings, analyst data and public sources linked in the text.
Axevil Capital, LLC is a company registered in the U.S. (State of Delaware, #6832739) with Exempt Reporting Adviser (ERA) status and is regulated by the U.S. Securities and Exchange Commission (SEC #802-126907). We work only with qualified investors. This material is for informational purposes only and does not constitute an offer, an investment recommendation or an invitation to enter into a transaction. Investments in private markets are illiquid, long-term and involve high risks, including the total loss of capital; past performance does not guarantee future results.
© Axevil Capital, 2026



