nextbig.dev
Vancouver, B.C. · — Intelligence on AI and the machines that run it
nextbig.dev
The Essay · Weekly

The week, argued.

One essay every week, built from the strongest of our daily calls: an original thesis you won't find in the source material, grounded in the week's reporting and closed with a falsifiable call. When the date comes due, we settle it in print.

Alongside the weekly essay, our evergreen field guides cover in depth the three beats we report every day: AI agents, GPU & infrastructure economics, and developer tools. The daily briefing turns the week's reporting into one falsifiable call, settled in public.

A broadsheet schematic: four labeled model cartridges feeding one router behind an interface bar, one cartridge pulled out and set on the bench, a red line switching the live traffic to the cheapest cartridge that clears the eval.
Dev July 9, 2026 11 min read

Make the Model Easy to Fire

Two frontier models shipped the same day, one at a quarter of the other's price. The durable skill in 2026 is not choosing an LLM. It is building so you can swap the one you chose in an afternoon. A builder's routing policy, eval harness, and the only cost number that matters.

Read →
A broadsheet cover: a stack of bars where the widest top bar is hollow and dashed while the narrow bottom bars are solid red, captioned money at the top, value at the bottom.
Infra July 5, 2026 13 min read

The Best AI Investment Doesn't Care Which Model Wins

In the first quarter of 2026, AI startups raised as much as they did in all of 2025, and two-thirds of it went to three companies. That concentration is the tell. The durable money in mid-2026 is not in the frontier labs everyone is crowding into. It is one layer down, in the model-agnostic scarcity every lab has to pay for no matter which one wins, ranked by how hard that scarcity is to build.

Read →
A schematic on warm paper: a dark central disc labeled harmful capability, ringed by a wide dashed red circle labeled safety margin. Hollow dots sit safely outside the ring, while three solid red dots, one marked 'you', are caught inside the margin band.
Security June 30, 2026 11 min read

The False Positive Was the Point

A safety classifier locked me out for trying to secure my own network. It read the words an attacker would use and could not see that the devices were mine. That false positive was not a glitch. Providers widen the safety margin on purpose, and this is how to do real security work without tripping it.

Read →
A severity title card: a single jailbreak prompt opening a small blast radius on a gated API and a large one on open weights, under the framework name Technique, Target, Terrain.
Security June 30, 2026 14 min read

Score the Blast Radius, Not the Prompt

The U.S. government pulled Claude Fable 5 for three weeks over a jailbreak nobody could agree was severe. Anthropic's answer is a four-axis severity framework drafted with its cloud resellers. It scores the attack and skips the two things that decide the damage: what is at stake, and where the model runs. Here is the framework that adds them back.

Read →
An editorial illustration in ink on warm paper: a long communal worktable with several empty chairs and, at its head, a smooth monolith in place of one chair standing for a non-human teammate, a single wire-red thread linking the seats, the whole table dwarfed by the vast hall that contains it.
AI June 25, 2026 13 min read

Anthropic Built the Best Coworker. Salesforce Owns the Channel.

Claude Tag puts a shared, remembering AI teammate inside Slack, and it is the best of its kind. The catch is the address: Slack belongs to Salesforce, which sells a rival coworker in the same channel. A scan of ten alternatives shows why the strongest product is rarely the one that wins the channel.

Read →
A border-style checkpoint with a long human queue waiting at an ID booth on one side and an empty fast lane beside it where a faceless automaton walks straight through carrying a ring of keys with no name tag.
Security June 22, 2026 6 min read

We Gave Agents Accounts Before Identities

Cloudflare hands agents live infrastructure in seconds while Claude makes humans verify who they are. That asymmetry is the next breach.

Read →
A fine technical schematic of an orbital computer: a small cube of server racks at the center, dwarfed by enormous deployable radiator panels fanning out on either side, a single solar wing above, thin red lines of heat escaping into empty space, with the curve of Earth below.
Infra June 20, 2026 12 min read

The Sun Is Free. The Cold Is Not.

SpaceX unveiled a 70-meter data center built to fly, Anthropic is shopping for gigawatts of orbital compute, and a startup already ran an H100 around the Earth. The number that decides whether any of it pays is not the one everyone is naming.

Read →
A vast server hall drawn in fine ink, its rows of machines fed not by a wall socket but by a single heavy power line that runs out the back of the building to a substation and a row of cooling towers on the horizon, the computer dwarfed by the grid that feeds it.
Infra June 20, 2026 12 min read

The Megawatt Is the Moat

The AI build-out spent three years treating compute as the scarce thing. The real shortage was power, and the grid behind the power. That reprices the whole stack: the unit of account becomes the megawatt, compute re-anchors to wherever the electricity is, and the next lasting advantage in AI infrastructure is an energy book.

Read →
A broadsheet schematic: one buyer ordering from both Nvidia and AMD, with a dashed equity line funding the smaller AMD node, the customer-funded second source.
Infra June 20, 2026 11 min read

The Second Source

AMD spent a decade as Nvidia's also-ran. In 2026 it became a credible alternative, and the reason says more about Nvidia's customers than about AMD's silicon.

Read →
A steep mountain ridge shaped like a falling cost curve, with a lone figure committing at the summit while others wait on lower ledges near a cheap plateau of server racks
Infra June 14, 2026 6 min read

Never Price a Model in Its First Month

Serving costs for a frontier model collapse by orders of magnitude in the six weeks after launch. Lock your stack in week one and you lock in the peak.

Read →
A vast, finely drawn machine alone in an empty hall, powered by a single thin cord that runs all the way across the floor to a heavy wall breaker lever caught mid-throw, the switch that controls it sitting far across the room.
Infra June 13, 2026 11 min read

The Kill Switch Was in the Contract

Washington pulled Anthropic's Fable 5 and Mythos 5 offline by federal letter on Friday, the first frontier model killed by directive. Export control just reached the deployment layer, and that changes what it is safe to build on.

Read →
A field guide cover for tracking the AI agent stack, set in the nextbig.dev broadsheet style
Dev June 13, 2026 8 min read

AI Agent News for Builders: Tracking the Agent Stack Without the Noise

A field guide to agent frameworks, orchestration, evals, and observability, and how to tell what ships from what only demos.

Read →
A briefing-playbook cover for GPU and AI infrastructure economics, set in the nextbig.dev broadsheet style
Infra June 13, 2026 9 min read

GPU & Infra Economics: A Briefing Playbook for ML Teams

How to read GPU supply, inference pricing, and datacenter economics, then decide where to run your models.

Read →
A builder's digest cover for the AI developer-tools stack, set in the nextbig.dev broadsheet style
Dev June 13, 2026 8 min read

AI Developer Tools News: Tracking the Devtools Stack for Builders

IDE assistants, agent SDKs, CI, evals, and observability: how to keep up with the tooling layer and know what's safe to adopt.

Read →
A field-guide cover for AI infrastructure news, set in the nextbig.dev broadsheet style
Infra June 13, 2026 8 min read

AI Infrastructure News: A Builder's Guide to the Compute Layer

What "AI infrastructure" actually means, the moves that matter, and how to follow GPU, datacenter, and inference news without the noise.

Read →
A market map plotting AI infrastructure players across model-routing ownership and capability governance, with an open opportunity zone in the top right
Infra June 9, 2026 16 min read

The AI Capability Control Plane

A market analysis of who controls AI execution today (model labs, gateways, security platforms, agent platforms, and identity), and the verified record showing every one of them converging on the same junction: the identity-aware request path.

Read →
Abstract visualization of AI agents as autonomous entities building digital structures
AI February 10, 2026 8 min read

Why AI Agents Are the New Startups

AI agents are starting to behave less like tools and more like companies. Most founders haven't caught up.

Read →
Developer workspace with multiple screens showing code, AI agent interfaces, and deployment dashboards
Dev February 8, 2026 10 min read

The Builder's Guide to Shipping with AI in 2026

Practical patterns, hard-won lessons, and the tools that actually matter for shipping products built on AI.

Read →
Data streams flowing through a neural filter, with signal amplified and noise fading away
AI February 5, 2026 7 min read

Signal vs. Noise: How We Curate Tech News with AI

How nextbig.dev's agent pipeline scores 300+ sources for builder relevance and drops the rest.

Read →
Vancouver skyline at dusk with digital nodes and connection lines overlaid, mountains in background
Startups February 3, 2026 7 min read

Vancouver's Quiet Rise as a Builder City

While the world watches Silicon Valley, a different kind of tech ecosystem is taking shape in the Pacific Northwest.

Read →
A digital lock being examined with a magnifying glass revealing hidden AI neural pathways underneath
Security January 30, 2026 9 min read

The Security Blind Spot in AI Apps

You secured your API. You hardened your database. But your AI features just opened attack vectors you haven't thought about.

Read →