# AI Sell-Off Is Noise and Demand Is Rising

Anthropic listed at $83 and NVIDIA guided to $130 billion as the Nasdaq had its worst week since 2020. In my view the NVIDIA headlines and the push against data centres are noise. Adoption keeps rising, and three profitable stocks in KXCO's AI ontology sit furthest below their analysts' targets.

Canonical HTML: https://www.livetradingnews.com/anthropic-is-trading-the-ai-sell-off-is-noise-and-demand-is-rising
Last modified: 2026-11-22

---

By Shayne Heffernan · 2026-11-22
Tags: $ANT, $NVDA, $ORCL, $BABA, Anthropic IPO, NVIDIA, AI stocks, AI ontology, data centres, data center moratorium, midterms, Claude Fable 5.2, GPT-6.2, Grok 4.7, Oracle, Alibaba, Ramp AI Index, Shayne Heffernan
Signed: ML-DSA-87, anchored on Armature L1.
Nothing in this article is investment advice.

In my view the AI sell-off of the past week is noise, and the business underneath it is accelerating. Anthropic priced a record-breaking initial public offering and began trading on the Nasdaq on 19 November 2026 ([Anthropic](https://www.anthropic.com/news/anthropic-prices-initial-public-offering); [NPR](https://www.stlpr.org/npr/2026-11-19/investors-scramble-for-a-piece-of-ai-giant-anthropic-in-blockbuster-ipo)). NVIDIA reported a quarter of $115.8 billion on 17 November 2026 and guided to $130 billion for the next ([NVIDIA](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27pr.htm)). The share of businesses paying for AI keeps rising, and their AI spending reached a new peak in October ([Ramp](https://ramp.com/data/ai-index-nov-2026)). Anthropic and OpenAI say the models they released this month do more for less money than the ones they replace ([Anthropic](https://www.anthropic.com/claude-fable-and-mythos-5-2); [OpenAI](https://openai.com/index/introducing-gpt-6-2-astra-sol/)). None of that changed in the week the market fell.

The market did fall, and hard. The Nasdaq Composite closed at 22,617.73 on Friday 20 November 2026, 10.0% lower on the week and 18.0% below its record close of 6 October 2026 ([Yahoo Finance](https://query1.finance.yahoo.com/v8/finance/chart/%5EIXIC?interval=1d&period1=1759276800&period2=1795392000)). NVIDIA closed at $125.53 on 20 November 2026, down 14.0% on the day and 22.9% on the week ([Yahoo Finance](https://query1.finance.yahoo.com/v8/finance/chart/NVDA?interval=1d&period1=1759276800&period2=1795392000)). That is 47.5% below its record close, also on 6 October 2026. Yahoo Finance's market statistician counted the Nasdaq's worst week since March 2020 and NVIDIA's worst since October 2008 ([Yahoo Finance](https://finance.yahoo.com/markets/live/stock-market-today-friday-november-19-dow-sp-500-nasdaq-anthropic-nvidia-141644188.html)).

The midterm elections have also given a louder voice to politicians who want to curtail the push into AI and data centres. In my view that push would prove disastrous for America's leadership of the world, and I set out why below.

This update draws on KXCO's AI ontology, the public record at [kxco.ai/ontology-live](https://kxco.ai/ontology-live), updated for this note with the news to 22 November and the analyst outlook at the 20 November close. The record reads the AI sector as typed relationships between companies, people, policies and supplies, each one dated and nearly all sourced, so a reader can check every link drawn here. It now holds 1,212 claims about 507 entities, 1,158 of them carrying a source ([KXCO AI record](https://kxco.ai/developers/blog/ontology-live-guide/)).

![Exhibit 1, OWL knowledge graph: the week's noise pressed on NVIDIA's share price, while its results, its customers and its suppliers kept moving, from KXCO's AI record](https://livetradingnews-media.nyc3.digitaloceanspaces.com/media/2026/11/22/cmpgg3-5e27f831c61583bb.svg)

Sources for Exhibit 1: [NVIDIA's results](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27pr.htm), [CFO commentary](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27cfocommentary.htm), [the resignation filing](https://www.sec.gov/Archives/edgar/data/0001045810/000104581026000082/nvidiaannouncesceotransiti.htm), [the SEC and California reports](https://finance.yahoo.com/markets/article/nvidia-stock-extends-25-trillion-rout-as-sec-probe-reportedly-investigates-its-accounting-194945274.html), [CNBC](https://www.cnbc.com/2026/11/20/nvidias-7-week-losing-streak-is-longest-since-2008-financial-crisis.html), [The Motley Fool](https://www.fool.com/investing/2026/11/20/here-is-why-nvidia-stock-crashed-on-friday/) and [the closes](https://query1.finance.yahoo.com/v8/finance/chart/NVDA?interval=1d&period1=1759276800&period2=1795392000).

## The noise, item by item

NVIDIA's co-founder Jensen Huang resigned as president, chief executive and director on 12 November, NVIDIA said in a [filing](https://www.sec.gov/Archives/edgar/data/0001045810/000104581026000082/nvidiaannouncesceotransiti.htm) with the SEC. On 17 November he said he would be honoured to return as chief executive if the board asked him to ([Fortune](https://fortune.com/2026/11/17/jensen-huang-honored-return-ceo-nvidia-board-approval-public-statement/)). The board's chair, Harvey Jones, said on the earnings call that the board had unanimously accepted the resignation and does not intend to reinstate him. NVIDIA has hired Heidrick & Struggles to find a permanent chief executive ([Yahoo Finance](https://finance.yahoo.com/technology/articles/nvidia-shuts-down-jensen-huang-093453815.html)). The board also set up a standing committee to oversee national security policy, export-control compliance and global trade risk ([NVIDIA](https://nvidianews.nvidia.com/news/nvidia-board-of-directors-establishes-national-security-and-export-control-oversight-committee)).

Then came the inquiries. The Wall Street Journal reported on 17 November that an SEC probe includes whether NVIDIA's disclosures misled investors. Reuters reported on 20 November that the probe is examining potential accounting issues, and the SEC declined to comment ([Yahoo Finance](https://finance.yahoo.com/markets/article/nvidia-stock-extends-25-trillion-rout-as-sec-probe-reportedly-investigates-its-accounting-194945274.html)). California's attorney general, Rob Bonta, served NVIDIA with an investigative subpoena on 19 November, investigating whether its advanced chips reached China and whether it misled investors ([Yahoo Finance](https://finance.yahoo.com/markets/article/nvidia-stock-extends-25-trillion-rout-as-sec-probe-reportedly-investigates-its-accounting-194945274.html)). Reuters itself noted that investigations are not evidence of wrongdoing and that many close without enforcement action ([The Motley Fool](https://www.fool.com/investing/2026/11/20/here-is-why-nvidia-stock-crashed-on-friday/)).

The selling spread from NVIDIA to the whole sector. Yahoo Finance reported that Friday's sell-off deepened after Anthropic's successful IPO failed to quell broad uncertainty about AI safety and financing. The Dow and the S&P 500 both closed in correction territory ([Yahoo Finance](https://finance.yahoo.com/markets/live/stock-market-today-friday-november-19-dow-sp-500-nasdaq-anthropic-nvidia-141644188.html)). AMD fell 20.5% on the week and TSMC 20%, and the S&P 500 had its worst week since April 2025 ([Yahoo Finance](https://finance.yahoo.com/markets/live/stock-market-today-friday-november-19-dow-sp-500-nasdaq-anthropic-nvidia-141644188.html)).

In my view each of these is either a question about governance at one company or a question of sentiment. None of them is a question about demand. A board can change its chief executive, and regulators can ask questions, without a single data centre being cancelled. The test is the business, and the business did not read the headlines.

## NVIDIA's business kept growing

NVIDIA reported revenue of $115.8 billion for the quarter to 25 October, up 103% on a year earlier, with Data Center revenue of $107.6 billion, up 110% ([NVIDIA](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27pr.htm)). It guided to $130.0 billion, plus or minus 3%, for the current quarter, which has an extra week. The guide assumes no Data Center compute revenue from China ([NVIDIA](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27pr.htm)). It has begun shipping its Vera Rubin NVL72 racks, which are running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius ([NVIDIA](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27pr.htm)). "Our momentum is real," said Colette Kress, the interim chief executive ([NVIDIA](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27pr.htm)).

The figure that says most about the future is on the supply side. NVIDIA's supply commitments rose to $390 billion from $279 billion a quarter earlier, mainly for memory bought over several years. The company says it has secured the components to meet demand for the next several years ([CFO commentary](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27cfocommentary.htm)). In my view a company does not commit $111 billion more to its suppliers in one quarter for demand it cannot see.

![Exhibit 2, chart: NVIDIA's daily closes, 1 October to 20 November 2026, with the record, the midterm fall, the resignation, the results and Friday's fall, against the average target](https://livetradingnews-media.nyc3.digitaloceanspaces.com/media/2026/11/22/cmpgg3-86d9ca1d605b8b97.svg)

The analysts held their rating. At the 20 November 2026 close, 63 analysts polled by S&P Global rated NVIDIA a consensus Buy with an average target of $244.18, 94.52% above the price ([Stock Analysis](https://stockanalysis.com/stocks/nvda/forecast/)). NVIDIA was worth $3.03 trillion, at 13.48 times trailing and 8.65 times forward earnings ([Stock Analysis](https://stockanalysis.com/stocks/nvda/statistics/)). KXCO's record rates NVIDIA's position critical: every major AI lab, cloud and sovereign programme depends on it, across 140 claims and 87 counterparties ([KXCO AI record](https://kxco.ai/ontology-live/)).

## Anthropic is now a public company

Anthropic priced 980 million Class A shares at $83 on 18 November 2026, and the shares began trading on the Nasdaq Global Select Market as ANT on 19 November 2026 ([Anthropic](https://www.anthropic.com/news/anthropic-prices-initial-public-offering)). The $83 price of 18 November 2026 was above the expected range of $70 to $78 ([Deadline](https://deadline.com/2026/11/anthropic-ai-volatile-ipo-shares-trading-nasdaq-market-1237229045/)). On 20 November 2026 Anthropic closed the offering with the underwriters' option of 147 million shares exercised in full, 1,127 million shares in all ([Anthropic](https://www.anthropic.com/news/ipo-close)). On 20 November 2026 it also closed a concurrent sale of 313,253,007 shares of Class N stock at the same price to AMD, Broadcom, Google, Micron, Microsoft, NVIDIA and TSMC Partners, for about $26.0 billion ([Anthropic](https://www.anthropic.com/news/ipo-close)).

At $83 a share that is about $93.5 billion from the offering and about $119.5 billion in all. NPR reported that the offering and placement valued Anthropic at around $1.5 trillion ([NPR](https://www.stlpr.org/npr/2026-11-19/investors-scramble-for-a-piece-of-ai-giant-anthropic-in-blockbuster-ipo)), and NBC News called it the largest IPO ever, topping the record SpaceX set a few months earlier ([NBC News](https://www.nbcnews.com/business/markets/live-blog/anthropic-ipo-live-updates-rcna610159)).

The shares opened at $111.50 on 19 November 2026, giving Anthropic a market value of about $2 trillion, and closed the first day at $97.17 ([NBC News](https://www.nbcnews.com/business/markets/live-blog/anthropic-ipo-live-updates-rcna610159)). On Friday 20 November 2026, the week's worst day, they closed at $88.91 ([Yahoo Finance](https://query1.finance.yahoo.com/v8/finance/chart/ANT?interval=1d&period1=1759276800&period2=1795392000)), still 7.1% above the offer price.

The list of buyers in the placement says more than the price. Google and NVIDIA each bought about $10.0 billion of stock, AMD about $2.5 billion, Broadcom, Micron and Microsoft about $1.0 billion each, and TSMC about $500 million, according to the final prospectus ([Anthropic prospectus](https://www.sec.gov/Archives/edgar/data/2130357/000162828026074944/anthropicpbc-424b4.htm)). These are the companies that sell Anthropic its chips, its memory, its manufacturing and its cloud. KXCO's record holds the other side of that ledger: Anthropic has committed to pay three of its backers at least $278.5 billion under agreements that run into the 2030s, whatever its usage ([KXCO AI record](https://kxco.ai/ontology-live/)). In my view a supplier that buys a customer's stock at the IPO price is telling you how much it expects that customer to buy.

![Exhibit 3, OWL knowledge graph: the seven companies that bought Anthropic stock in the placement, and the computing Anthropic has committed to buy from them, from KXCO's AI record](https://livetradingnews-media.nyc3.digitaloceanspaces.com/media/2026/11/22/cmpgg3-80bdeb169dd16965.svg)

Sources for Exhibit 3: [the prospectus](https://www.sec.gov/Archives/edgar/data/2130357/000162828026074944/anthropicpbc-424b4.htm), [the closing release](https://www.anthropic.com/news/ipo-close) and [KXCO's AI record](https://kxco.ai/ontology-live/).

Customers are voting the same way. Ramp's AI Index, which tracks what American businesses pay for, found that 44.8% of the businesses on its platform paid for Anthropic in October, up 1.1 percentage points, while OpenAI held at 40.6% ([Ramp](https://ramp.com/data/ai-index-nov-2026)).

## What the ontology shows that a news feed does not

A news feed reports each of these events once. KXCO's AI ontology holds them as typed links between the same companies, so the connections appear in one view. The placement is a case in point. The record holds the seven share purchases beside the computing Anthropic has committed to buy from the same companies, and its finding on circular deals counts 21 vendor and customer pairs in which money flows back to the company that funded it ([KXCO AI record](https://kxco.ai/ontology-live/)). Exhibit 3 is drawn from those links.

The four knowledge graphs in this note are drawn from one OWL 2 knowledge graph built for it, with 20 classes, 15 object properties and 36 individuals. OWL, the Web Ontology Language, is a family of languages for authoring ontologies, and the World Wide Web Consortium announced its current version, OWL 2, in 2009 ([Wikipedia](https://en.wikipedia.org/wiki/Web_Ontology_Language)). An automated OWL reasoner finds the graph consistent and refuses each of three planted statements that break it, such as an event recorded as buying stock. It also states facts no single link records: Anthropic is an agent, because an AI lab is a company, a company is an organisation and an organisation is an agent.

The record also keeps its history. When Anthropic priced, the record did not overwrite the roadshow claim of 10 November. It closed that claim at 18 November and recorded the pricing as the claim that supersedes it ([KXCO AI record](https://kxco.ai/ontology-live/)). A reader can see what was expected, what happened and when the record changed. In my view that is what institutions need from AI analysis: every figure dated, sourced and open to challenge, and nothing rewritten after the fact.

## The midterms gave the curbs a louder voice

Democrats won control of both chambers of Congress on 3 November, according to an NPR analysis of race calls by the Associated Press ([NPR, via OPB](https://www.opb.org/article/2026/11/04/2026-midterms-democrats-win-the-senate/)). AI was on the ballot. An NBC News exit poll found that 56% of voters rated AI and data centres at least a major factor in their vote for Congress ([Time](https://time.com/article/2026/11/04/midterm-elections-results-ai-industry-data-centers/)). Voters sided against data centres in every one of 41 local ballot measures in 10 states except one, in Harwood, North Dakota ([CNBC](https://www.cnbc.com/2026/11/04/data-center-ballot-measures-2026-midterm.html)). Annie Andrews, who called for a two-year moratorium on new projects, beat a candidate the data-centre and accelerationist super PACs had backed ([Time](https://time.com/article/2026/11/04/midterm-elections-results-ai-industry-data-centers/)).

The winners have kept up the pressure. Senator Bernie Sanders answered the vote with a call for a national data-centre moratorium, now ([CNBC](https://www.cnbc.com/2026/11/05/democrats-data-center-restrictions-midterms.html)). On 17 November Representative Jim McGovern called outside the Capitol for a national moratorium on new AI data centres. Two incoming members, Adam Hamawy of New Jersey and Will Lawrence of Michigan, joined the call ([Missouri Independent, via KTTN](https://www.kttn.com/house-democrats-urge-national-moratorium-on-new-ai-data-centers/)). The same day Loudoun County, Virginia, voted to pause action on data-centre and electrical substation applications for up to 12 months, which could affect 33 active applications ([Loudoun County](https://www.loudoun.gov/m/NewsFlash/Home/Detail/10993)).

Washington has limits on what the curbs can do. President Trump keeps his veto, and before the vote he said he would block Democrats' legislation if they took Congress: "I'll become Mr. Veto" ([NPR, via OPB](https://www.opb.org/article/2026/11/04/2026-midterms-democrats-win-the-senate/)). The Senate passed a bipartisan permitting bill 62 to 33 on 19 November that aims to speed up reviews for power plants and transmission lines and to make data centres pay for the energy they use. It now goes to the House ([AP, via ABC News](https://abcnews.com/US/wireStory/senate-backs-bill-speed-permitting-reviews-force-data-138308460)). Harvard's Ash Center expects the near-term action at state level, on permitting, utility rates and tax incentives. It expects Congress to start with oversight hearings ([Ash Center](https://ash.harvard.edu/articles/what-the-2026-midterm-results-mean-for-technology-policy/)).

In my view the push to curtail AI and data centres, for whatever unfathomable reason it is being made, would prove disastrous for America's leadership of the world. KXCO's record puts the United States at roughly 75% of global GPU-cluster performance against about 15% for China ([KXCO AI record](https://kxco.ai/ontology-live/)). It also records that China is building a frontier AI stack of its own, with a capability gap of about 2.7% between the best American and Chinese models as of March 2026 ([KXCO AI record](https://kxco.ai/ontology-live/)). In my view America's lead sits in its chips, its data centres and its power, China is already close on the models, and every data centre the United States declines to build narrows the one lead China has not yet closed.

![Exhibit 4, OWL knowledge graph: the post-election pressure on US data centres, the checks on it, and the US lead in AI computing that the build-out protects, from KXCO's AI record](https://livetradingnews-media.nyc3.digitaloceanspaces.com/media/2026/11/22/cmpgg3-29eb43dd547d0cda.svg)

Sources for Exhibit 4: [NPR](https://www.opb.org/article/2026/11/04/2026-midterms-democrats-win-the-senate/), [Time](https://time.com/article/2026/11/04/midterm-elections-results-ai-industry-data-centers/), [CNBC](https://www.cnbc.com/2026/11/04/data-center-ballot-measures-2026-midterm.html), [KTTN](https://www.kttn.com/house-democrats-urge-national-moratorium-on-new-ai-data-centers/), [Loudoun County](https://www.loudoun.gov/m/NewsFlash/Home/Detail/10993), [AP](https://abcnews.com/US/wireStory/senate-backs-bill-speed-permitting-reviews-force-data-138308460) and [KXCO's AI record](https://kxco.ai/ontology-live/).

## In the real world, adoption is accelerating

Ramp's index put business adoption of AI at 56.8% of businesses in October, up 1 percentage point, slightly faster than September's 0.66-point gain, and AI spending rose to a new peak ([Ramp](https://ramp.com/data/ai-index-nov-2026)). That is one measure. KXCO's record tracks six independent ones, taken by different methods of different populations, and they move the same way. The share of employed US adults using generative AI for work rose from 33.3% in the third quarter of 2024 to 45.2% in the second quarter of 2026 on the St. Louis Fed's survey. OpenAI says ChatGPT reaches 1.2 billion people a week. Alphabet's model interfaces process nearly 32 billion tokens a minute, more than four times a year earlier ([KXCO AI record](https://kxco.ai/ontology-live/)). The record notes that Ramp's customers skew to smaller, tech-forward firms, and that the direction survives that caveat.

Demand is contracted as well as used. The four largest cloud sellers carry about $2.48 trillion of contracted backlog, revenue that customers have committed to and that has not yet been recognised ([KXCO AI record](https://kxco.ai/ontology-live/)). Amazon, Alphabet, Meta and Microsoft plan roughly $730 billion of AI capital spending in 2026, up 78%, a pace the record flags as faster than the revenue behind it ([KXCO AI record](https://kxco.ai/ontology-live/)). In my view NVIDIA's $390 billion of supply commitments is the chipmaker buying ahead for that spending ([CFO commentary](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27cfocommentary.htm)).

In my view demand is strong and it is not about to change. A business that has put AI into its workflows does not take it out because a share price fell, and the contracts behind the build-out run for years.

## The models keep advancing

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on 9 November 2026, two versions of the same underlying model ([Anthropic](https://www.anthropic.com/claude-fable-and-mythos-5-2)). On Anthropic's figures of 9 November 2026 they generate output up to 30% faster than the 5.1 models, at a lower price of $8 per million input tokens and $40 per million output tokens ([Anthropic](https://www.anthropic.com/claude-fable-and-mythos-5-2)). In Anthropic's testing, reported on 9 November 2026, a team of 1,000 Fable 5.2 agents fully formalised the proof of Fermat's Last Theorem in 30 hours. The first formalisation, by an unreleased research version of Claude, took 11 days ([Anthropic](https://www.anthropic.com/claude-fable-and-mythos-5-2)). Fable 5.2 also solved two open problems in theoretical physics, both concerning black holes ([Anthropic](https://www.anthropic.com/claude-fable-and-mythos-5-2)).

OpenAI released GPT-6.2 Astra, its new frontier model for the hardest projects, and GPT-6.2 Sol, its model for everyday work, on 17 November ([OpenAI](https://openai.com/index/introducing-gpt-6-2-astra-sol/)). Astra's prices on OpenAI's application programming interface (API) are 20% lower than GPT-6 Astra's ([OpenAI](https://openai.com/index/introducing-gpt-6-2-astra-sol/)). OpenAI says Astra scores 5.2 points higher than GPT-6 Astra on Agents' Last Exam, a test of multistep professional work, at less than a third of the cost ([OpenAI](https://openai.com/index/introducing-gpt-6-2-astra-sol/)).

SpaceXSI, the former xAI, released Grok 4.7, its frontier model for coding, agentic tasks and knowledge work, on its API on 21 September 2026 ([SpaceXSI](https://docs.x.ai/developers/release-notes)). It has a 500,000-token context window and costs $2 per million input tokens and $6 per million output tokens below 200,000 prompt tokens ([SpaceXSI](https://docs.x.ai/developers/release-notes)). It is the fastest riser in business adoption from the smallest base: 6.3% of Ramp businesses paid SpaceXAI, as Ramp calls it, in October, up 2.2 percentage points in a month ([Ramp](https://ramp.com/data/ai-index-nov-2026)).

![Exhibit 5, OWL knowledge graph: the three labs, the models they released, and the share of businesses paying each one, from KXCO's AI record](https://livetradingnews-media.nyc3.digitaloceanspaces.com/media/2026/11/22/cmpgg3-b32365094ea135ac.svg)

Sources for Exhibit 5: [Anthropic](https://www.anthropic.com/claude-fable-and-mythos-5-2), [OpenAI](https://openai.com/index/introducing-gpt-6-2-astra-sol/), [SpaceXSI](https://docs.x.ai/developers/release-notes) and [Ramp](https://ramp.com/data/ai-index-nov-2026).

In my view the pattern matters more than any one benchmark. Each release does more work per dollar, and cheaper capability widens the market. That is how adoption compounds, and it is why the demand behind the build-out keeps rising while the share prices fall.

## Undervalued on the analysts' targets: three stocks from KXCO's AI ontology

KXCO's record carries an analyst outlook for 19 listed AI companies, restamped for this note to the 20 November 2026 close. It holds each company's price, its analysts' consensus target and its trailing price-to-earnings (P/E) multiple, and the gap between price and target is the measure of value used here. At the 20 November 2026 close the average upside to the consensus target is 74%, up from 51% at the 13 November 2026 close ([KXCO AI record](https://kxco.ai/ontology-live/)). I ranked the companies in that outlook that earned a profit over the past twelve months by their upside. Three lead the list.

![Exhibit 6, chart: upside to the analysts' target at the 20 November close for the 19 companies in KXCO's AI outlook, with trailing P/E, and the three profitable leaders](https://livetradingnews-media.nyc3.digitaloceanspaces.com/media/2026/11/22/cmpgg3-b11ec1775a0b8efa.svg)

**Oracle (NYSE: ORCL).** Oracle closed at $101.19 on 20 November 2026 against a consensus target of $236.18, 133.40% higher ([Stock Analysis](https://stockanalysis.com/stocks/orcl/forecast/)). It trades at 15.87 times trailing and 11.85 times forward earnings ([Stock Analysis](https://stockanalysis.com/stocks/orcl/statistics/)). KXCO's record shows a backlog of $664 billion at 31 August against a fiscal 2027 revenue guide of at least $90 billion, more than seven years of revenue already under contract ([KXCO AI record](https://kxco.ai/ontology-live/)). In my view the market is pricing Oracle's capital spending and ignoring the contracts that pay for it.

**Alibaba (NYSE: BABA).** Alibaba closed at $81.08 on 20 November 2026 against a consensus target of $162.28, 100.15% higher, with 40 analysts rating it a Strong Buy ([Stock Analysis](https://stockanalysis.com/stocks/baba/forecast/)). It trades at 11.48 times trailing and 9.57 times forward earnings ([Stock Analysis](https://stockanalysis.com/stocks/baba/statistics/)). The record shows cloud growth of 45% in its latest quarter and a HK$80 billion share placement with every dollar committed to AI ([KXCO AI record](https://kxco.ai/ontology-live/)).

**NVIDIA (Nasdaq: NVDA).** NVIDIA closed at $125.53 on 20 November 2026, and the analysts' average target of $244.18 is 94.52% above it ([Stock Analysis](https://stockanalysis.com/stocks/nvda/forecast/)). It trades at 13.48 times trailing earnings ([Stock Analysis](https://stockanalysis.com/stocks/nvda/statistics/)), after a quarter of $115.8 billion and a guide of $130 billion ([NVIDIA](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27pr.htm)). In my view a company growing revenue at 103% does not belong at 13 times earnings.

The next names on the same screen are Baidu, at 90%, and TSMC, at 84% ([KXCO AI record](https://kxco.ai/ontology-live/)). Cerebras, SpaceX and Intel are loss-making over the past twelve months, so the screen leaves them out. The targets are the analysts' consensus, which KXCO's record carries and dates. The ranking and the calls are mine.

## What to watch

- NVIDIA's quarterly report on Form 10-Q, which was still to be filed when its results came out on 17 November 2026 ([CFO commentary](https://www.sec.gov/Archives/edgar/data/1045810/000104581026000088/q3fy27cfocommentary.htm)).
- The House vote on the Senate's permitting bill ([AP, via ABC News](https://abcnews.com/US/wireStory/senate-backs-bill-speed-permitting-reviews-force-data-138308460)).
- Anthropic's first weeks of trading against the $83 offer price of 18 November 2026 ([Anthropic](https://www.anthropic.com/news/anthropic-prices-initial-public-offering)).

The noise will pass, as it always does. In my view the build-out it distracted from will not. The record behind this note is open to read at [kxco.ai/ontology-live](https://kxco.ai/ontology-live), and KXCO's work on the ontology of markets is at [kxco.ai](https://kxco.ai).

Stocks mentioned: $ANT (NASDAQ), $NVDA (NASDAQ), $ORCL (NYSE) and $BABA (NYSE).

Shayne Heffernan, Ph.D., is the founder of the KnightsBridge Group ([shayneheffernan.com](https://www.shayneheffernan.com)).

---

This Markdown mirrors https://www.livetradingnews.com/anthropic-is-trading-the-ai-sell-off-is-noise-and-demand-is-rising. The HTML page is canonical.
Site index for AI clients: https://www.livetradingnews.com/llms.txt
