Outcome sizes are exponentially larger than historical norms; $1B valuation may be the new Series A price with same blunt multiple to $20B outcomes; picking is harder due to volume but conv…

Hardware/physical AI has longer time to validation (moving atoms not bits) — can't measure zero-to-$1M ARR but zero-to-working-prototype. Requires more capital and time, driving more fund c…

Sequoia data across decades shows best investments always from sponsors with highest conviction, not consensus. SpaceX had a 1/10 vote but sponsor forced partner visits. Airbnb was controve…

Sequoia's 5-decade data shows best fund returns always from deals where sponsor had highest conviction, even if controversial (SpaceX vote of 1); process optimizes for conviction not agreem…

Across decades of Sequoia funds, the best returns consistently come from deals where the sponsoring partner had highest conviction; increasing ownership percentages mathematically never bea…

New AI-native markets (greenfield) grow vertically for 3-5 years until saturation, then become replacement markets with normal growth. Current exponential growth in coding/support is greenf…

Sequoia's historical analysis shows best returns always come from highest-conviction sponsors, not from optimizing ownership percentages. Partnership encourages fierce debate and 'front-ste…

Agent traffic already at parity with human traffic (Cloudflare data); in 5 years will be 1000x. This creates a parallel agent economy requiring bits-perfect platforms (not pixel-perfect UIs…

Cloudflare data shows agent traffic will exceed human traffic 1000x in 5 years; companies must build for agent decision-making (biases, pre-training data) not human UI, creating new categor…

As agents become primary information consumers, a new category AEO emerges to optimize visibility in chat interfaces; Profound exemplifies this; the agent economy will be as large as the hu…

Agents can swap menial tools instantly, but databases and enterprise systems retain switching costs via data gravity, compliance, and trust built over time; these moats persist regardless o…

Agents will swap commoditized tools instantly, but enterprise infrastructure (databases, ERP) retains high switching costs from data gravity, compliance, and trust built over time. Enterpri…

Enterprises won't switch databases or core systems quickly despite agent buying; data gravity, compliance, and trust built over time remain moats, so infrastructure and implementation servi…

Sequoia invests in both infrastructure (Fireworks, ClickHouse) and applications (Rill) because Jevons paradox makes building cheaper and increases total demand; infrastructure wins now, app…

Application winners are transient; infrastructure winners (inference, database, observability) compound durably as all app-layer companies build on them. Sequoia invests in both layers but…

Fireworks (inference) and Rill (finance automation) can both be massive because infrastructure wins now during exponential build-out while applications compound later into sticky real-econo…

Fireworks (inference), ClickHouse (database), Mercor (recruiting infra) are durable bets — whoever wins at app layer uses this infrastructure. Human brain bad at exponentials but also at ho…

Making software building cheaper (via AI) increases total demand for software, leading to more software engineers hired than ever; human creativity expands to fill the new capacity, creatin…

AI dramatically lowers software creation cost, triggering Jevons paradox — total software output and engineer hiring increase because more ideas become economically viable. Data shows recor…

Legal market broad enough for horizontal platform (Harvey) and vertical specialists (Solve in patent law); both Harvey and Legora founders invested in Solve, validating unbundling thesis.

Brain-computer interfaces (BCI) is the most underfunded category — that's where all the smart kids are going now, like ML/AI 10 years ago. 10-year horizon for blossoming. Following talent f…

Following talent migration patterns (ML/AI 10 years ago), brain-computer interfaces now attract the smartest researchers; long-horizon but high-conviction next platform shift.

Top technical talent is moving to brain-computer interfaces, signaling the next major platform shift. Following talent flows has historically identified transformative categories early (ML/…

When AI reaches 500 IQ (vs ~120 today), it will discover cures for diseases like the guest's father's neurological condition; Demis Hassabis and DeepMind focusing on biology signals the fie…

Superintelligent AI (500 IQ) will discover cures for neurological and chronic diseases. DeepMind's focus on biology and London's concentration of life-sciences talent position this as the h…

Current models at ~120 IQ but exponential progress toward 500 IQ will enable scientific breakthroughs (curing neurological diseases, biology); Sequoia investing in life sciences frontier co…