Daily Briefing — April 16, 2026

Yet another layoff story. Snap just cut around 1,000 jobs, roughly 16% of its entire workforce, and the official reason handed to staff was "rapid advancements in artificial intelligence." That framing is doing a lot of heavy lifting. The actual trigger was pressure from Irenic Capital Management, an activist investor who wrote directly to CEO Evan Spiegel last month demanding cost cuts and a reduced headcount. AI is the narrative. Investor pressure is the mechanism.

This is worth paying attention to because the "AI made us do it" explanation is becoming a standard move across the industry. Microsoft, Amazon, Oracle, Block, and now Snap have all leaned on some version of this story while cutting tens of thousands of roles. The honesty level varies wildly. Sometimes AI genuinely is reshaping workflows. Sometimes a company just needs to show Wall Street a leaner cost structure and AI gives the announcement a forward looking spin that "we need to cut costs" does not.

Nvidia built a moat so wide that for years, the answer to "who controls AI infrastructure" was basically just "them". The company does not just make the best chips. It also makes the software that tells those chips what to do, and that combination has been almost impossible to compete with.

A startup called Wafer is using AI to automate one of the most specialized and valuable skills in the entire tech stack: writing kernel code, the low level software that tells hardware exactly how to execute instructions as efficiently as possible. Right now, that skill lives in the heads of a very small number of engineers who get paid accordingly. Wafer is trying to make that skill accessible to anyone with a model and a chip.

The downstream effect is significant. If optimizing code for custom silicon becomes something an AI agent can do well, then the advantage of owning Nvidia specific expertise shrinks. Companies running Google TPUs, Amazon Trainium, or Apple silicon could close the gap faster. The democratization of chip level programming is a story about where specialized knowledge becomes a commodity and what replaces it.

Google just dropped a dedicated Gemini app for Mac, and the headline feature is an Option + Space shortcut that pulls up a floating chat bubble from anywhere on your desktop. You can share a window, let Gemini read what is on screen, and ask it questions without switching contexts. If that sounds familiar, it should. It is the same basic pattern that Anthropic's Claude for Mac and OpenAI's ChatGPT desktop app have been refining for months, and that Apple itself is pushing with its upgraded Spotlight.

The interesting piece is what Google's app does not do yet. Claude and ChatGPT have started moving beyond answering questions toward actually performing actions on your computer, clicking, typing, navigating apps on your behalf. Gemini on Mac is still stuck on the "answer questions about what you are looking at" rung of that ladder. That is a meaningful gap in a market where the race is no longer about who has the smartest chatbot but about who can become the default interface layer on your desktop.

The broader pattern here is that the operating system itself is quietly turning into a battleground. Four companies (Apple, Google, Anthropic, OpenAI) are all trying to own the keyboard shortcut that pulls up your AI assistant, and the prize is becoming the thing you reflexively reach for dozens of times a day.

Amazon just announced an $11.57 billion deal to acquire Globalstar, the small satellite company whose low-Earth-orbit network quietly powers the emergency SOS feature on your iPhone and Apple Watch. As part of the same announcement, Amazon said it is partnering with Apple going forward. Translation: Apple's off-grid emergency communication is about to be running on top of Amazon's expanding satellite ambitions, not some neutral third party.

The strategic story here is Starlink. Elon Musk's satellite network has been running away with the low-Earth-orbit market, and Amazon's Project Kuiper (now being rebranded to Project Leo) has been chasing from behind since 2023. Buying Globalstar gets Amazon an existing fleet of 24 satellites, valuable spectrum rights, and a deep relationship with Apple in one move. It is the kind of acquisition that instantly changes the shape of a race.

Satellite connectivity is quietly becoming critical plumbing for AI. Edge devices, autonomous systems, and always-online agents all need bandwidth in places where cell towers do not reach. The companies that own the sky-to-ground pipes are positioning themselves to collect rent on the next generation of connected hardware. That is why a company like Amazon is willing to write an eleven-figure check for what looks, on the surface, like a niche satellite operator.