StocksEquities, earnings & valuations
Panmure Liberum's Klement finds AI-stock analyst ratings close to dot-com-peak levels
Joachim Klement finds analyst ratings on key AI stocks nearly as one-sided as tech ratings at the 2000 peak. Meanwhile, a Nasdaq correction, Jensen Huang's resignation and Nvidia's doubling revenue send conflicting signals about how the AI boom might end.
From the Klement on Investing article of November 18, 2026 and cited sources: Klement puts 83% of ratings on key AI stocks at Strong Buy and 2% at Sell. [1] [2]
SOURCES[1] What the last two bubbles tell us about the AI boom –… klementoninvesting.substack.com[2] What the last two bubbles tell us about the AI boom –… theglobeandmail.com
AI-assisted, not human-edited. Methodology Corrections Who owns this site
- Original source
- Klement on Investing article, independent perspective
- Brief by
- AT Pulse research desk AI-assisted; publisher and contact on the desk page
- When
- November 18, 2026 article published
- Covered
- Nov 18 edition
- Evidence
- 6 sources across 6 sites
- 1What the last two bubbles tell us about the AI boom –…klementoninvesting.substack.com
- 2What the last two bubbles tell us about the AI boom –…theglobeandmail.com
- 3Stocks plunge as midterm results send AI trade into a…localnews8.com
- 4Nvidia CEO Jensen Huang resigns amid spy scandal – Al…aljazeera.com
- 5Jensen Huang resigns as Nvidia CEO – Washington Postwashingtonpost.com
- 6NVIDIA Announces Financial Results for Third Quarter…nvidianews.nvidia.com
StrongHow it’s measured
Substack · 9 likes · 2 comments
Reach blends public engagement (views, likes, comments, shares, stars) from the platforms where the story or its original source lives: YouTube, Hacker News, GitHub, Substack and similar. A video and the article written from its transcript share one score, taken from the video. Platforms with no data are left out rather than counted as zero, and the age of a story never lowers its score.

Klement puts 83% of ratings on key AI stocks at Strong Buy and 2% at Sell. That compares with 85% and 0% for tech favourites at the 2000 peak, and 53% Buy and 12% Sell for U.S. banks before the 2008 crisis. Near-unanimous bullishness did not warn of either earlier decline. [1] [2]
Klement describes two ways a boom can end. In 2000 the Nasdaq rose 95% in about five months, then fell 32% in three weeks: a 'bang.' In 2007 bank stocks rose only 12% and deflated slowly: a 'whimper.' Goldman Sachs' AI basket, up 51% from April to June then down 24% in July and 17% in October, falls between the two. Klement does not predict which ending is coming. [2]
The column follows a rough month. On Nov 5 the Nasdaq fell 5.52% into correction, Nvidia fell 9.3% and the chip index fell 12.3%, amid worries that midterm results, including a Texas governor who backed a data-center moratorium, would slow AI buildouts. On Nov 12 Jensen Huang resigned as Nvidia CEO, and CFO Colette Kress became interim chief. [3] [4] [5]
analyst sentiment is near 2000-peak levels while Nvidia's reported revenue doubled.
Read the full assessment
Implication: consensus ratings have not warned of past declines, so leaders planning AI infrastructure should weigh data-center policy, supplier governance and valuation risk alongside headline growth.
Executive brief
Analyst ratings on AI stocks are about as one-sided as they were on tech stocks at the 2000 peak. Joachim Klement of Panmure Liberum writes that 83% of ratings on key AI stocks are "Strong Buy" and 2% are "Sell". At the 2000 peak the figures for tech favourites were 85% and 0%. Before the 2008 crisis, U.S. bank ratings were 53% Buy and 12% Sell (Globe and Mail/Reuters). His column came out after a hard month for AI stocks: a Nasdaq correction, and Nvidia's CEO resigning. A day before it appeared, Nvidia reported revenue up 103% from a year earlier. Prices and analyst sentiment are moving one way while reported earnings move another.
What changed and event timeline
AI basket slides again
Klement cites Goldman Sachs' U.S. Broad AI basket. It rose 51% from April to June, fell 24% in July and lost another 17% in October ().
Nasdaq enters correction
The Nasdaq fell 5.52%, Nvidia 9.3% and the chip index 12.3%. The trigger was worry that Democratic midterm wins, including a Texas governor who campaigned for a data-center moratorium, would slow AI buildouts ().
Jensen Huang resigns
He resigned after reports of an affair with a suspected Chinese intelligence operative and a relationship with a subordinate. CFO Colette Kress became interim CEO (;).
Nvidia reports record quarter
Q3 FY2027 revenue was $115.8B, up 103% year on year. Data-center revenue was $107.6B. Q4 guidance is $130B ±3% ().
Klement's Reuters column
He compares analyst sentiment today with 2000 and 2008 and asks whether the AI boom ends "with a bang or a whimper" ().
Capabilities and access
No model or product launch is involved; this is a story about financial markets. The one product signal is Nvidia's Rubin platform, which the company says is "off to an explosive start" (NVIDIA). Nvidia's Q4 outlook leaves out China data-center compute revenue, so access to China remains outside its forecast. The Reuters original is paywalled; the full text is syndicated by the Globe and Mail.
Technical analysis for researchers and developers
Klement's method compares the mix of analyst ratings at two past market peaks with the mix now, alongside price paths (Globe and Mail/Reuters): The column does not say which stocks make up "key AI stocks" or which data vendor supplied the ratings, so the numbers cannot be reproduced. Three reference points do not support statistical inference.
Read the full section
Klement's method compares the mix of analyst ratings at two past market peaks with the mix now, alongside price paths (Globe and Mail/Reuters):
- 2000: the Nasdaq rose 95% from October 1999 to March 2000, then fell 32% in three weeks. That is the "bang" pattern.
- 2007: bank stocks rose only 12% between 2006 and February 2007 and deflated slowly. That is the "whimper" pattern.
- Now: AI stocks sit between the two.
The column does not say which stocks make up "key AI stocks" or which data vendor supplied the ratings, so the numbers cannot be reproduced. Three reference points do not support statistical inference.
Claims and evidence
- Ratings are lopsided (83% Strong Buy; Nvidia 80% Buy, one Sell among 85 analysts).
- Nasdaq correction and chip slump on Nov 5 are confirmed by independent reporting (CNN via KIFI).
- Nvidia's revenue and guidance come from Nvidia's own release; they are reported results, not independent analysis (NVIDIA).
Read the full section
- Ratings are lopsided (83% Strong Buy; Nvidia 80% Buy, one Sell among 85 analysts). This is Klement's own analysis and has not been independently checked (Globe and Mail/Reuters).
- Nasdaq correction and chip slump on Nov 5 are confirmed by independent reporting (CNN via KIFI).
- Nvidia's revenue and guidance come from Nvidia's own release; they are reported results, not independent analysis (NVIDIA).
- Nvidia's market value fell below $4T, its lowest earnings multiple in 14 years (Al Jazeera).
Context and prior work
Klement's frame matches a wider debate. On Nov 5, Truist's Keith Lerner called the drop a "knee-jerk reaction" to the election. Ed Yardeni said a slower AI buildout was welcome because the boom had been "too frenzied and speculative" (CNN via KIFI). His point about analysts staying bullish into 2000 and 2008 is a historical observation, not a forecast (Globe and Mail/Reuters).
Limitations, safety and contested findings
- Ratings are bubble-like, but Nvidia's reported revenue doubled (NVIDIA). The two signals point in opposite directions.
- The selloff had political, governance and valuation causes. Separating them is interpretation, not established fact.
- Klement compares only two earlier episodes, and he does not say which outcome will happen.
Read the full section
- Sentiment versus fundamentals. Ratings are bubble-like, but Nvidia's reported revenue doubled (NVIDIA). The two signals point in opposite directions.
- Mixed causes. The selloff had political, governance and valuation causes. Separating them is interpretation, not established fact.
- Small sample. Klement compares only two earlier episodes, and he does not say which outcome will happen.
- Governance risk. The Huang case involved national-security questions, with investigations by the FBI, Congress and Nvidia's board (Al Jazeera).
Business and practitioner implications
- Rules on data centers are now a direct risk to AI-infrastructure plans. Texas is the named case (CNN via KIFI).
- Leadership change at the main AI-chip supplier is a supply-chain and key-person risk that buyers can track.
- Read analyst consensus carefully. Klement's data shows that near-unanimous bullish ratings have not warned of past declines.
Read the full section
- Data-center policy. Rules on data centers are now a direct risk to AI-infrastructure plans. Texas is the named case (CNN via KIFI).
- Supplier concentration. Leadership change at the main AI-chip supplier is a supply-chain and key-person risk that buyers can track.
- Read analyst consensus carefully. Klement's data shows that near-unanimous bullish ratings have not warned of past declines.
- No investment view. Nothing here is investment advice.