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/COATUE/Charts of the Day + Public Marke…
NEWS
// NEWSLETTER ISSUE
COATUE

Charts of the Day + Public Markets Update Deck

DATE October 7, 2026SOURCE COATUEPARTICIPANTS COATUE MANAGEMENT
AI AgentsAI InfrastructureAI SafetyDeep LearningLarge Language ModelsVenture Capital
In this episode
AnthropicOpenAINvidiaMicrosoftAmazonMetaCoatue ManagementBroadcomTSMC
// SUMMARY

Coatue — Charts of the Day + Public Markets Update Deck
Coatue — Charts of the Day + Public Markets Update Deck

Coatue — Charts of the Day + Public Markets Update Deck (2)
Coatue — Charts of the Day + Public Markets Update Deck (2)

Coatue — Charts of the Day + Public Markets Update Deck (3)
Coatue — Charts of the Day + Public Markets Update Deck (3)

Coatue — Charts of the Day + Public Markets Update Deck (4)
Coatue — Charts of the Day + Public Markets Update Deck (4)

1. Key Themes

Theme: AI is a new "Age," not just another technology "wave"

Coatue reframes the unit of analysis from waves to ages

The email's central claim is a change in how to size and think about the AI opportunity: it's not the next bump in a series of tech cycles, but a civilizational-scale shift.

  • "We are moving from 'waves' to 'ages.'"
  • Slide title: "Moving from waves to ages"
  • Slide title (the old framework): "How we used to think about innovation"

The old "wave" framework showed AI breaking the pattern of successive, bounded cycles

The first chart plots mainframe (1960–1980), PC (1980s), networking (1990s), desktop internet/Web 1.0 (2000s), mobile internet/Web 2.0 (2010s), and cloud/SaaS (2015–2020) as sequential bell curves, each larger than the last. AI (2022–2026+) is drawn as a dotted blue line that does not peak like the others but shoots upward.

  • "AI" is labeled on the dotted blue trajectory with the span "2022 - 2026+" and no visible peak.
  • Each prior wave is labeled with its era (e.g., "Mobile Internet (Web 2.0)", "Cloud / SaaS", "Desktop Internet (Web 1.0)").

Each successive age has been dramatically shorter, implying acceleration

The "ages" chart sets out the durations and what each age accomplished: the Agrarian Age (10,000 BCE–1750) lasted "12,000 years" and "Fed the world"; the Industrial Age (1750–1950) lasted "200 years" and "Built the world"; the Information Age (1950–2020) lasted "70 years" and "Digitized the world." The Intelligence Age (2020+) is marked with a "?" for duration and is described as "Automate the world."

  • "12,000 years — Fed the world"
  • "200 years — Built the world"
  • "70 years — Digitized the world"
  • "? — Automate the world"

Theme: The unit of value is shifting from software to tokens, and from paperwork to work itself

Tokens as the new economic primitive

Coatue's summary line for each age shows a progression of what gets displaced: "Farms replaced foraging," "Machines replaced muscle," "Software replaced paperwork," and now "Tokens replace work." This frames AI not as a productivity tool for knowledge workers but as a substitute for labor broadly.

  • "Tokens replace work"
  • The Intelligence Age's building blocks are listed as "GPUs," "Token factories," and "Transformers."

The AI stack is full-stack, spanning silicon through models

The wave chart's AI column groups chip/infrastructure and model players together, suggesting the opportunity is distributed across the stack rather than concentrated in one layer.

2. Contrarian Perspectives

Don't benchmark AI against prior tech cycles, because it's a different category of event

Most investors map AI onto the cloud/SaaS or mobile playbook and ask where we are in the "cycle." Coatue argues this mental model understates the opportunity. The chart shows prior waves as symmetrical, mean-reverting bell curves that rise and fall, while AI is drawn as a curve that doesn't turn over.

  • "We are moving from 'waves' to 'ages.'"
  • Chart title: "How we used to think about innovation" (framing the wave model as the outdated view).
  • Evidence: the age durations compress from "12,000 years" to "200 years" to "70 years," with AI's duration left as "?", implying it may be both shorter-cycled and larger in scope than anything prior.

Value capture may sit in "tokens" and infrastructure rather than in traditional software

The framing "Software replaced paperwork... Tokens replace work" implies software, the dominant investable category of the last age, is itself the thing being superseded. Note that Salesforce, Microsoft, and Amazon appear in the Cloud/SaaS curve, while the AI column is led by semiconductor and model names (TSMC, NVIDIA, Broadcom, OpenAI, Anthropic).

  • "Tokens replace work"
  • AI-column grouping places "tsmc," "nvidia," and "BROADCOM" at the top alongside model builders.

3. Companies Identified

IBM / HP — Mainframe-era incumbents. Mentioned as representative of the 1960–1980 wave. Quote: logos shown under "Mainframe 1960-1980."

Dell / Intel / HP — PC-era players. Mentioned as representative of the 1980s wave. Quote: logos shown under "PC 1980s."

Ericsson / Nokia / Cisco — Networking-era players. Mentioned as representative of the 1990s wave. Quote: logos shown under "Networking 1990s."

Facebook / Google / Amazon — Desktop internet (Web 1.0) leaders. Mentioned as representative of the 2000s wave. Quote: logos shown under "Desktop Internet (Web 1.0) 2000s."

Apple / Tencent / ByteDance — Mobile internet (Web 2.0) leaders. Mentioned as representative of the 2010s wave. Quote: logos shown under "Mobile Internet (Web 2.0) 2010s."

Salesforce / Microsoft / Amazon — Cloud/SaaS leaders. Mentioned as representative of the 2015–2020 wave. Quote: logos shown under "Cloud / SaaS 2015 - 2020."

TSMC — Semiconductor foundry. Mentioned as an AI-era beneficiary. Quote: logo shown in the AI column under "2022 - 2026+."

NVIDIA — GPU/accelerator maker. Mentioned as an AI-era beneficiary, consistent with the "GPUs" and "Token factories" building blocks of the Intelligence Age. Quote: logo shown in the AI column; "GPUs" listed under "Intelligence Age 2020+."

Broadcom — Custom silicon and networking. Mentioned as an AI-era beneficiary. Quote: logo shown in the AI column under "2022 - 2026+."

Meta — Social platform and AI model/infrastructure investor. Mentioned as an AI-era player. Quote: logo shown in the AI column under "2022 - 2026+."

Microsoft — Appears in both the Cloud/SaaS and AI columns. Mentioned as a platform that carried leadership across eras. Quote: logos shown in both "Cloud / SaaS" and "AI" columns.

Amazon — Appears in the Desktop Internet and Cloud/SaaS columns, notably absent from the AI column. Mentioned as a multi-era winner. Quote: logos shown under "Desktop Internet (Web 1.0)" and "Cloud / SaaS."

OpenAI — Frontier AI lab. Mentioned as an AI-era leader. Quote: logo shown in the AI column under "2022 - 2026+."

Anthropic — Frontier AI lab. Mentioned as an AI-era leader. Quote: logo shown in the AI column under "2022 - 2026+."

Note: Coatue's disclaimer states logos "reflect illustrative examples of companies within the applicable sector or trend listed, are not necessarily past or current Coatue investments."

4. People Identified

No individual people are named in this email.

5. Operating Insights

Frame your company's opportunity as labor substitution, not software tooling

If "tokens replace work" is the thesis of the age, product positioning built around replacing or automating units of work (rather than assisting a human user) aligns with the dominant economic narrative. Coatue's progression from "paperwork" to "work" is the pitch to emulate.

  • "Software replaced paperwork" → "Tokens replace work"

Think in terms of the full AI stack when deciding where to build

Coatue's AI column spans foundry, chips, networking silicon, hyperscalers, and model labs, and the age's building blocks include "GPUs," "Token factories," and "Transformers." Operators should map which layer their business depends on (compute supply, model access, distribution) and where margin accrues.

  • "GPUs," "Token factories," "Transformers" listed as the Intelligence Age's components.

Plan for compressed cycle times

With ages shrinking from "12,000 years" to "200" to "70," planning horizons built on 5–10 year technology cycles may be too long. Build optionality and speed into roadmaps.

  • "70 years — Digitized the world," followed by "? — Automate the world."

6. Overlooked Insights

The AI column drops several previous-era winners

Amazon appears in two prior waves and Microsoft in the Cloud/SaaS and AI columns, while Salesforce, Apple, Google, Tencent, and ByteDance do not appear in the AI column. This suggests Coatue sees leadership turnover, with prior-wave dominance not guaranteed to carry forward.

  • AI-era logos: "tsmc," "nvidia," "Meta," "BROADCOM," "Microsoft," "OpenAI," "ANTHROPIC."

Cloud/SaaS overlaps with the AI era in the chart

The AI curve begins around "2015 - 2020" on the timeline, rising from the base of the cloud/SaaS peak and crossing its declining side, implying AI emerged while the prior wave was still at its height rather than after it faded.

  • The dotted AI line starts near the "2015 - 2020" label and runs to "2022 - 2026+."