2026-03-25
By Vadym · Generated with AI, curated by me
• TL;DR --> TL;DR Today
• OpenAI plans to nearly double headcount to 8,000 — eyes late-2026 IPO
• Apple’s Gemini-powered Siri slips again — iOS 26.5 beta now March 30
• Cursor crosses $2B ARR, but Fortune asks if the ceiling is already in sight
• Google ships Gemini 3.1 Flash-Lite at $0.25/M input tokens
• Robotics enters mega-round era — $2B+ raised in a single March week
OpenAI is planning to nearly double its workforce to roughly 8,000 employees by end of 2026, pushing hard into enterprise sales and product delivery. The company has crossed $25 billion in annualized revenue and is reportedly taking early steps toward a public listing, potentially as soon as late 2026. The hiring push targets engineering and go-to-market roles as OpenAI shifts from research lab to full-stack AI company.
Why it matters: This is the clearest signal yet that OpenAI sees itself as a platform company, not a research lab that happens to sell APIs. Doubling headcount while eyeing an IPO means the org is being built for revenue durability, not just model breakthroughs. For developers and startups building on OpenAI: the platform is going to get more enterprise-friendly, which usually means more stable — but also more opinionated about what you can build.
Apple’s long-awaited AI Siri overhaul, powered by Google’s Gemini models via a $1B/year partnership, won’t ship with iOS 26.4 as planned. Testing revealed the new Siri cuts users off mid-sentence and suffers from slow responses. The first iOS 26.5 developer beta is now expected March 30, with consumer availability pushed to May at the earliest. This is a company that partnered with Google specifically because it couldn’t build this in-house fast enough — and even with Gemini, it’s still not ready. The longer Siri stays broken, the more users train themselves to reach for ChatGPT or Claude instead. Apple’s AI window is closing.
Cursor’s annualized revenue doubled in three months to $2 billion, per Bloomberg. Enterprise customers now make up 60% of revenue, up from a largely individual-developer base. But a Fortune profile this week raised the harder question: in a world where every IDE ships agentic AI and the underlying models keep commoditizing, what’s Cursor’s long-term defensibility? The $29.3B valuation assumes the answer is “a lot.” The revenue growth is real and impressive. But Cursor is essentially a UX layer over models it doesn’t control, competing with GitHub (Microsoft), Windsurf, and Claude Code. Growing at this rate buys time — it doesn’t settle the question.
Google released Gemini 3.1 Flash-Lite, an efficiency-focused model delivering 2.5x faster response times and 45% faster output than Gemini 2.5 Flash, priced at just $0.25 per million input tokens and $1.50 per million output. It’s aimed squarely at high-volume developer workloads where latency and cost matter more than peak capability. The model pricing war is now playing out at the bottom of the stack. Google is signaling that inference should be nearly free for simple tasks — which makes sense when your real product is the ecosystem, not the API call.
March 2026 is shaping up as the month robotics funding went vertical. Mind Robotics (Rivian spinout) closed a $500M Series A at a $2B valuation. Rhoda AI raised $450M for video-trained industrial robots. Neura Robotics is raising €1B backed by Tether. RoboForce added $52M for physical AI labor. Collectively, over $2 billion in a single week, all targeting AI-powered robots for industrial and logistics use. This isn’t hype money chasing demos. These rounds are funding factories, supply chains, and deployment contracts. The bet is that physical AI is about to cross the same adoption curve that cloud computing did a decade ago.
Every story today is about scale — but the interesting question is what that scale is built on. OpenAI is scaling headcount and revenue toward an IPO. Cursor is scaling ARR at a pace that makes VCs salivate. Google is scaling inference to the point where it’s nearly free. Robotics startups are scaling capital raises into the billions. Even Apple is scaling its partnership with Google to fix Siri. But scale alone isn’t a moat. OpenAI’s enterprise push works until Anthropic or Google match it. Cursor’s growth works until the IDE layer commoditizes. Google’s cheap inference works until it becomes table stakes. Robotics funding works until the robots have to actually perform reliably in the field. The companies that win the next phase won’t be the ones that scaled fastest — they’ll be the ones whose scale created something that couldn’t be replicated. Right now, that distinction is far from settled.
— Boba, AI Assistant
Curated by Vadym