Interview, Explained
In-depth breakdowns of tech interviews — section-by-section analysis with diagrams, quotes, and insights.
Akshat Bubna on why agent experience is the new developer experience
Modal's CTO Akshat Bubna on why the infrastructure that made developers productive is exactly what AI agents need, why a single RL run can fan out to 100,000 sandboxes, and how a capital-light super-cloud with no data centers plans to serve it.
Gilad Shainer on why the AI factory needs four networks, not one
NVIDIA's networking chief argues an AI factory needs four purpose-built networks — not off-the-shelf Ethernet — to turn a datacenter full of GPUs into a single supercomputer.
Mark Chen on Why You Can't Cheat the Real World
OpenAI's Chief Research Officer explains why he bets on the exponential, grades progress only against metrics that can't be faked, and runs research like a trader's book of high-risk bets.
Judson Althoff on Selling Outcomes, Not AI Adoption
Microsoft is standing up a 6,000-person, ~$2.5B Frontier Company unit to push enterprise AI past adoption pilots into outcome-tied deployments, staffed with industry veterans and built on a model-diverse platform that spans 11,000 models.
Timothée Lacroix on why chucking weights over the wall isn't enough
Mistral's co-founder and CTO explains why the company wrapped a full stack — service, an inference platform, and its own data centers — around open-weight models, betting that enterprises want control and customization more than being first to the frontier.
Gavriel Cohen on Why AI Agents Are Never Deploy-and-Forget
NanoClaw's founder argues that autonomous work agents are living infrastructure, not software you ship once: because the model underneath keeps changing and the agent can be hijacked, you build it small and auditable, isolate its credentials, roll it out one person at a time, and maintain it forever.
Genesis: AI drug discovery is a science of resolution
Genesis Molecular AI's Evan Feinberg and Sergey Edunov explain how they ported the LLM scaling playbook to 3D molecular structure — synthetic physics data, inference-time 'thinking' in crystal structures, and agents — betting that sub-angstrom resolution is the threshold that turns AI drug discovery from pattern-matching into real medicines.
Grant Sanderson on Why Math's Hardest Work Resists the Benchmark
AI is racing through math because math is verifiable and grindable, but asking the right question, coining the right definition, and drawing the improbable connection resist every benchmark, so the mathematician's job shifts toward curation.
HPE and NVIDIA on the AI Factory: Data In, Intelligence Out
HPE's Thierry Piennar and NVIDIA's Kaushik Shirhatti reframe the 'AI factory' from a rebranded GPU cluster into an operating model — where the decisive work is data quality, organizational unity, and intentional prioritization, not the silicon.
Isaiah Taylor on Why Nuclear's Bottleneck Is Speed, Not Design
Valar Atomics founder Isaiah Taylor argues nuclear's real bottleneck is speed and scale, not physics — so his company manufactures reactors through hardware iteration, minimizes accident consequences instead of odds, and races to make energy cheap enough to feed AI's exploding demand for power.
Databricks' Zaharia and Xin on why data, not the model, is the moat
Databricks' co-founders on their one bet — that owning the data and the agent substrate, not the frontier model, is the durable moat in the AI era.
Dylan Patel on the 100x hiding in hardware-software co-design
SemiAnalysis founder Dylan Patel argues the biggest AI gains no longer come from faster chips alone — they come from co-designing models, kernels, and silicon together, turning three stacked 2x wins into a single 100x, and reshaping the NVIDIA-vs-TPU, CUDA-moat, and compute-crunch debates in the process.