Interview, Explained

In-depth breakdowns of tech interviews — section-by-section analysis with diagrams, quotes, and insights.

Decoded·

Ankit & Francois on Why AGI Needs a World Model, Not More Data

Ankit & Francois· Hosts of Y Combinator's Decoded

World models — learning to predict the next state of the world — may be the route to sample-efficient AI, especially in robotics where brute-force reinforcement learning is intractable.

ReasoningTrainingRoboticsAI Infrastructure
Build Mode·

Sydney Sykes on why strategic fit wins partnerships

Sydney Sykes· Global VC Alliances and Partnerships lead at NVIDIA

Sydney Sykes, who runs Nvidia's VC alliances, breaks down how startups win corporate partnerships — by aligning to priorities, not by having the best tech — and how founders should build a cap table of realists and dreamers.

AI CompanyBusiness StrategyAI Infrastructure
Y Combinator·

Matthieu Rouif & Eliot Andres on Why Ambition Is a Trainable Skill

Matthieu Rouif & Eliot Andres· Co-founders of PhotoRoom (CEO & CTO)

PhotoRoom's founders argue that ambition is not a personality trait but a skill you train: reset your benchmark against bolder peers, aim at targets so high they pull the company up to meet them, shrink every experiment to a V0 that AI can now ship in days, and earn scale by going deep on one thing at a time.

AI CompanyBusiness StrategyMultimodal
Equity Podcast·

Matt Murphy on why Anthropic's harness beat the model

Matt Murphy· Partner at Menlo Ventures

Menlo's Matt Murphy explains Anthropic's stunning revenue leap as less about model quality than the harness around it — Claude Code, MCP, and Skills — plus a VC willing to pay outcome-level prices to back the inevitable strong number two, in a wave that spreads product-led with almost no friction.

AI CompanyBusiness StrategyAgentsLLM
NVIDIA·

Jensen Huang on the Factory That Turns Electricity Into Intelligence

Jensen Huang· Founder and CEO of NVIDIA

In a fireside chat at Wistron's U.S. plant, Jensen Huang recasts data centers as AI factories that turn electricity into intelligence tokens, argues this new industrial layer will become as fundamental to society as agriculture and railroads, and ties it to reindustrializing American manufacturing.

AI InfrastructureGPUInferenceBusiness Strategy
Yahoo Finance·

Lisa Su on Why Inference Is AI's Real Inflection Point

Lisa Su· Chair and CEO of AMD

AMD's Lisa Su argues AI has crossed from training to inference — the everyday running of models — and that meeting accelerating demand will require abundant compute, open ecosystems, and capacity committed 12 to 24 months ahead.

AI InfrastructureInferenceAgentsBusiness Strategy
No Priors·

Andy Fang & Stanley Tang on Why DoorDash's Moat Is the Real World

Andy Fang and Stanley Tang· Co-founders of DoorDash

DoorDash's co-founders explain why they treat the company as an AI and robotics operator, building a custom delivery robot and a natural-language ordering agent from the customer's use case backward and defending it with ten billion deliveries of real-world data.

AgentsRoboticsBusiness Strategy
Startup School Paris·

Stanislas Polu on Why No Single AI Lab Will Win

Stanislas Polu· Co-founder of Dust; former OpenAI engineer

Dust co-founder Stanislas Polu argues the durable AI company is the one that never bets on a single frontier lab — staying model-agnostic, horizontal, and defended by network effects rather than scaffolding.

LLMAgentsAI CompanyBusiness Strategy
CNBC·

AMD's Helios Bets the AI Market Is Too Big to Be Nvidia's Alone

Forrest Norrod & Vamsi Boppana· Data center and AI leaders at AMD

AMD unveils Helios, its first rack-scale AI system, as an open, full-stack challenge to Nvidia, betting that a market Nvidia controls more than 95% of is expanding fast enough for a real second source to win.

GPUAI InfrastructureInferenceBusiness Strategy
The a16z Show·

Qasar Younis & Peter Ludwig on Why Physical AI's Moat Is the Real World

Qasar Younis & Peter Ludwig· Cofounders of Applied Intuition

Applied Intuition's cofounders argue physical AI's real bottleneck is never model access but the friction of the physical world — data you must go collect, safety you must prove, latency you must beat — and that friction is exactly what becomes the moat.

RoboticsAgentsAI InfrastructureBusiness Strategy
Latent Space·

Bo Wang & Ci Chu on why virtual cells need causal data, not atlases

Bo Wang & Ci Chu· SVP, Biomedical AI & SVP, AI-Enabled Discovery, Xaira Therapeutics

Xaira's virtual-cell leads on why the bottleneck to predicting biology isn't the model but the data — and how industrializing CRISPR perturbation screens into causal datasets let a diffusion model predict how cells respond to genes it has never seen.

LLMTrainingAI Company
Bloomberg Tech·

Kai-Fu Lee on Why the Money Is Above the Model

Kai-Fu Lee· CEO of 01.AI

Kai-Fu Lee argues that as models grow smarter than all humans and open source narrows the gap, the value in AI moves off the model and onto applications that change a company's bottom line, and 01.AI is betting on hands-on enterprise deployment in markets its US rivals are unlikely to enter.

LLMOpen SourceBusiness StrategyAI Company