Chamath returns with $135M; Rocket Lab buys Iridium

The AI coding startup 8090 Labs secured a $135 million Series A led by Salesforce Ventures, positioning its enterprise‑grade Software Factory to automate production‑quality code. Palihapitiya, its founder, announced he will now run the company as CEO, betting the AI rush will reshape software development like social media did for his career.

As venture firms balloon into asset‑manager scale, they lose the ability to fund $5‑50 M companies, creating a permanent split between trillion‑dollar capital seekers and specialized investors chasing fragmented mid‑market opportunities. Founders must turn to smaller, agile capital sources, reshaping deal flow and rewarding firms that can stay nimble.

Rocket Lab announced a definitive agreement to acquire Iridium for $8 billion in cash and stock, merging its launch and satellite‑manufacturing chops with Iridium's 66‑satellite L‑band network and spectrum. The deal gives Rocket Lab an instant platform for IoT, direct‑to‑device and PNT services, positioning it as a full‑stack rival to SpaceX and Amazon in the telecom arena.

Eddie Kim, Gusto’s CTO, walked a five‑person squad through a 10‑week sprint that produced a full AI‑powered payroll line using Claude Code, a permanent Zoom room, and a ‘trash‑can’ PR workflow, skipping design tools, Jira, and docs. The case shows how tiny teams can ship complex AI products fast.

Tom Tunguz explains that many startups launch without an immediate moat and must earn it through scale, brand, or network effects. He separates leading moats (built at founding) from lagging moats (earned over years) with examples like Salesforce and Snowflake, urging founders to be honest about their moat status and focus on execution.

Eight months after launching its paid evaluation service, Arena announced a $100 million annualized revenue run rate. The milestone shows AI labs are paying for human‑judged benchmarking, turning a free research leaderboard into a must‑have infrastructure for model development.

AI agents are becoming a hidden distribution channel, favoring open-source, API-first tools that they can read, call, and stitch together. Companies like Supabase, Resend, PostHog and n8n see ten-fold user growth because agents can auto-configure and recommend their services. This shifts growth strategy from product-led to agent-led.

Governor Newsom's deal with Anthropic lets every state and local agency use Claude at half price, adds free workforce training and technical assistance. The move aims to accelerate AI‑driven productivity in government while keeping deployment responsible, giving California a head start on AI‑enabled public services.

AI lowers the minimum viable team size, letting companies do more with fewer people and dodge bloated meetings. Miguel Carranza explains how RevenueCat’s OCTO squad used AI to let a small group of senior engineers move fast, handling routine code while preserving talent and speeding up product bets.

A ten‑second attention threshold makes even order‑of‑magnitude speedups irrelevant if users still have to wait. Colin Breck shows how performance gains that look impressive on paper can fail to change workflows, and why engineers must chase the human limits that truly drive impact.

OpenTag lets teams run self‑hosted AI agents inside Slack, mirroring Claude‑like capabilities without lock‑in. Built on CopilotKit’s open SDK, it supports any LLM and custom tools, giving developers full control over runtime and data. This open‑source approach could accelerate enterprise adoption of in‑workspace AI assistants.

Moondream warns that the GPU market is entering a correction as supply outpaces shifting AI demand. The post breaks down oversupply signals and how they will squeeze AI infrastructure margins, urging operators to rethink capacity planning before prices plunge.

Databricks co‑founder Arsalan Tavakoli warns that any B2B software monopoly will crumble within 12‑24 months as AI budgets surge and firms blindly chase token spend. He says real AI value hinges on context, not models, meaning companies that tie spend to clear outcomes will dominate the next wave.

A U.S. Census Bureau working paper shows firms that spent $30 per employee per month on AI grew headcount 10.2% and entry‑level staff 12%, refuting the claim AI wipes out junior jobs. The gains are concentrated in large, data‑rich companies, hinting at uneven labor‑market impacts.