Teahose.
SIGN IN
NEW HERE — WHAT TEAHOSE DOES
We read the entire AI & tech firehose — so you don't have to.
PODPodcastsAll-In, No Priors, Acquired…
NEWNewslettersStratechery, Newcomer…
PAPPapersPhysical AI research
PHProduct Huntdaily launches
VCInvestor ScoutSequoia, a16z, Benchmark…
CLAUDE DISTILLS →
7 reads, 30 sec each — free, 6 AM ET.
+ a live graph of the companies, people & themes underneath.
HOME/DATA DRIVEN VC/🔥AI Labs Asked to Slow Down Whi…
NEWS
// NEWSLETTER ISSUE
DATA DRIVEN VC

🔥AI Labs Asked to Slow Down While NVIDIA Keeps Shopping - Here's the State of the Market in September 2026

DATE September 20, 2026SOURCE DATA DRIVEN VCPARTICIPANTS ANDRE RETTERATH
In this episode
// SUMMARY

1. Key Themes

Macro tightening meets private-market froth — two different clocks

Central banks resumed hiking while private markets kept absorbing record capital, creating a widening disconnect between public and private conditions.

"The Fed announced a rate increase from 3.75% to 4% on 16 September, its first hike in three years, and signalled another." "Private markets are on a different clock." "August brought $42 billion private money into 1,500 startups according to Crunchbase, seven of them raising a billion or more."

AI capital concentration is extreme and narrowing further

Exit and public-market activity in venture-backed tech is being dominated by a single name, while AI labs themselves are voicing caution about the pace of development even as capital keeps flowing.

"US venture-backed tech has raised close to $90 billion in public money this year, 83% of it SpaceX." "Dario Amodei called for slower frontier development and Sam Altman agreed. In the same month, NVIDIA paid 86x revenue for Hugging Face."

M&A: revenue multiples expanding while EBITDA multiples compress

Buyers are increasingly paying for growth/platform value rather than current profitability — a shift in how strategic and PE buyers are underwriting deals.

"deal value climbs past $2.8T, revenue multiples up while EBITDA multiples ease" "Buyers paid more of the top price and less of the extreme one, and they paid it on revenue rather than earnings."

Insurance brokerage — a quiet consolidation hotspot

An unglamorous sector is quietly absorbing a disproportionate share of large-cap M&A activity, suggesting a structural roll-up theme investors may be missing amid AI headlines.

"Insurance brokerage is the theme nobody was watching. Three of the ten largest deals sit in the sector: USI to Aon at $17,000M, Baldwin to Sequence Holdings at $7,700M, and Steadfast Group to KKR and Dragoneer at $5,500M, spanning both strategic and private equity buyers."

Public market rerating is concentrated at the very top

The best-performing/highest-quality software names are pulling further away from the rest of the pack, rather than a broad-based multiple expansion.

"Compared to August, the top of the market pulled away. Everything else stood still." "Top 10 average 21.0x, median 15.5x, up from 17.4x and 14.5x in August. Overall median at 2.2x, from 2.1x."


2. Contrarian Perspectives

  • The NVIDIA-Hugging Face deal signals a genuine platform acquisition, not just talent buying — against the common assumption that mega AI M&A is mostly acquihire-driven.

"Hugging Face to NVIDIA at 86x is the outlier and the largest strategic AI acquisition in the dataset this year. It is also the first time a top-multiple deal has also been a top-size deal, which is what a genuine platform purchase looks like as opposed to an acquihire."

  • Public market cap gains mask a top-heavy, narrow rally — a caution against reading broad equity strength (S&P +12%) as evidence of a healthy or diversified market.

"Equities held, with the S&P 500 up roughly 12% this year." Yet: "The whole top cluster rerated and Cloudflare simply rerated less."

  • The AI industry publicly asking to slow down while simultaneously paying record multiples for acqui­sitions is a sign of internal contradiction, not caution — the author frames the juxtaposition itself as noteworthy tension worth watching rather than taking the "slow down" rhetoric at face value.

"Then the AI trade was questioned from inside. Dario Amodei called for slower frontier development and Sam Altman agreed. In the same month, NVIDIA paid 86x revenue for Hugging Face."


3. Companies Identified

  • Anthropic — Frontier AI lab. Mentioned as the top private company by valuation and for delaying its IPO twice.

"Anthropic holds first at $965B... Anthropic pushed its IPO to November, a second delay after October slipped. It filed confidentially on 1 June and is reportedly weighing a new model release ahead of the listing."

  • OpenAI — Frontier AI lab. Mentioned for ruling out a 2026 listing on safety grounds.

"OpenAI is out for the year. Sam Altman ruled out a 2026 listing on safety grounds in mid-September."

  • NVIDIA — Semiconductor/AI infrastructure giant. Mentioned for its record-multiple acquisition of Hugging Face and subsequent market reaction.

"NVIDIA paid 86x revenue for Hugging Face. Semiconductors fell almost 6% in a session."

  • Hugging Face — AI model/tools platform. Mentioned as the target of NVIDIA's landmark acquisition.

"The widely covered Hugging Face to NVIDIA deal follows at $12,900M."

  • Oura — Wearable health tech company. Mentioned for filing publicly for IPO.

"Oura filed publicly on 3 September under the ticker OURA, though it is not yet on the IPO calendar for this month."

  • Palantir — Data/software company. Mentioned as retaking the top valuation multiple spot with standout growth and margins.

"Palantir retakes first at 37.5x from 29.2x, reversing last month's fall on 56% NTM revenue growth with a 63% EBITDA margin. No other name in the table pairs growth and margin at that level."

  • CrowdStrike, Cloudflare — Cybersecurity/infrastructure companies. Mentioned for rerating alongside Palantir at the top of the multiples table.

"CrowdStrike climbs to 34.8x from 28.0x. Cloudflare rose to 33.2x from 30.6x and still lost the top spot, which is the useful detail."

"Four names entered: Palo Alto Networks at 20.5x, VeriSign at 15.3x on the best margins in the table, JFrog at 14.5x and Cadence at 11.4x."

  • Aon / USI Insurance Services — Insurance brokerage. Mentioned as the largest M&A deal of the month.

"USI Insurance Services to Aon at $17,000M leads, 5.7x revenue and 14.5x EBITDA."

  • ByteDance, Databricks, Stripe, Waymo, Revolut, Reliance Retail, Anduril — Large private companies. Mentioned in the Top 10 private market valuation table.

"ByteDance sits third at $480B... Reliance Retail returns at eighth with $101B after dropping out last month. Anduril enters at tenth with $61B."

  • SpaceX — Aerospace company. Mentioned as dominating venture-backed public capital raises this year.

"US venture-backed tech has raised close to $90 billion in public money this year, 83% of it SpaceX."

  • Standard Metrics — Portfolio data/AI analyst platform for VCs. Newsletter sponsor, mentioned as case study for AI-driven portfolio management.

"Standard Metrics builds that foundation for 150+ investment firms. We collect, extract, and verify portfolio data, then make it available to query in our AI Analyst."


4. People Identified

  • Dario Amodei — CEO of Anthropic. Mentioned for publicly calling for slower frontier AI development.

"Dario Amodei called for slower frontier development and Sam Altman agreed."

  • Sam Altman — CEO of OpenAI. Mentioned for agreeing with slowdown calls and ruling out an IPO this year.

"Sam Altman ruled out a 2026 listing on safety grounds in mid-September."

  • Andre Retterath — Author/investor, writer of Data Driven VC. Mentioned as newsletter host and data curator.

"Hi, I'm Andre and welcome to my newsletter Data Driven VC which is all about becoming a better investor with data and AI."


5. Operating Insights

  • Watch multiple divergence, not just direction: When a "top cluster" rerates together, the story isn't the single winner but the shared repricing — track relative movement within a peer set, not just the leader, to understand what's really driving valuation (e.g., quality/growth premium vs. sector-wide sentiment).

"This is not a Palantir story. The whole top cluster rerated and Cloudflare simply rerated less."

  • Treat "platform vs. acquihire" as a diagnostic for M&A quality: A deal that is simultaneously top-multiple AND top-size signals genuine strategic/platform intent rather than a disguised talent grab — useful framework when evaluating comparable transactions or your own M&A exit positioning.

"It is also the first time a top-multiple deal has also been a top-size deal, which is what a genuine platform purchase looks like as opposed to an acquihire."

  • Revenue multiples over EBITDA multiples signal buyer conviction in growth over current profitability — operators positioning for exit should understand which metric buyers are currently anchoring on.

"Buyers paid more of the top price and less of the extreme one, and they paid it on revenue rather than earnings."


6. Overlooked Insights

  • Average deal size is falling even as total deal value climbs, suggesting volume/breadth of M&A activity is increasing even as headline mega-deals grab attention — a nuance easily missed when focusing only on aggregate deal value.

"Average deal size fell to $231M from $252M in August and $265M in July. It is now $2M above 2015 and $44M to $103M above each of the past five years."

  • Global rate hikes are synchronized across major economies, not just the U.S., with Japan hitting a three-decade high — a macro signal with implications for global cost of capital that could get lost under the AI-dominated narrative.

"The ECB and Bank of Japan moved in the same week. Japan increased rates to 1.25% from 1%, a 31-year high."