The AI economy is splitting in two, and this week showed exactly where the fault line runs.

Atlassian cuts 10%, Block axes 4,000, and Meta's Avocado misses its own bar. Meanwhile Tesla grows headcount, Nvidia preps inference chips with OpenAI locked in as a customer, and Alibaba ships agentic AI to consumers in minutes. The companies winning aren't just deploying AI faster. They're rebuilding around it entirely.

Andrew Yang's labor tax pitch sounds radical until you realize Tesla and Nvidia are already living in the world he's describing.

TLDR

• Nvidia GTC 2026: Jensen Huang faces four pressure tests this week: inference chips, Rubin Ultra, geopolitics, and a robotics wildcard

• Alibaba OpenClaw: Alibaba's JVS Claw app lets non-technical users spin up agentic AI in minutes, days after Baidu shipped its own client

• Meta Avocado delay: Meta pushed its Avocado flagship past May after internal benchmarks fell short of Google's bar

• Yang's automation tax: Andrew Yang calls for shifting the US tax burden from labor to automation beneficiaries as AI threatens half of entry-level white-collar jobs

• Tesla grows headcount: Musk adds staff while rivals cut, betting Optimus robots push output-per-human to what he calls "nutty high"

• AI co-scientists: Google DeepMind's Gemini co-authored 42% of an AMOC climate study and cut peer review time 10x

• SaaS SBC crisis: Falling valuations and AI revenue pressure force companies like Atlassian to cut headcount and shrink stock-based compensation before dilution compounds

Nvidia GTC 2026 – Four big questions Jensen Huang must answer

Inference, Rubin Ultra, geopolitics, and a robotics wildcard on the table

What's happening: Nvidia kicks off GTC 2026 this week with Jensen Huang expected to unveil new inference-focused chips, next-gen Rubin Ultra systems, and a roadmap that addresses growing competitive pressure 🚀

The details:

• Inference chip: Nvidia is reportedly readying an inference-focused product incorporating Groq technology, with OpenAI as a key customer. The design leans on SRAM over HBM, sidestepping tight high-bandwidth memory supply 🧠

• Rubin Ultra: Nvidia's next-gen Rubin platform demands far more power than prior generations. Analysts watch whether hyperscalers commit to supporting it and what the Feynman generation copackaged optics architecture delivers next

• Robotics revenue: Nvidia logged roughly $6 billion in robotics-related revenue last quarter and is calling for an aggressive humanoid timetable, making it a near-term rather than long-term story ⚙️

• Geopolitics: Nvidia halted H200 production for China and shifted capacity to Rubin. US export restrictions could make Nvidia a gatekeeper for global AI chip sales, with Saudi Arabia and UAE filling the demand gap

Why this matters: CUDA's grip weakens in inference, meaning competitors can poach Nvidia customers without needing to crack its training ecosystem 🎯

Read more: Business Insider

Alibaba drops OpenClaw app, Baidu follows: Alibaba's "JVS Claw" app lands on iOS and Android this week, letting non-technical users deploy OpenClaw agentic AI within minutes, days after Baidu shipped its own Android OpenClaw client. The apps feed China's "raising lobsters" craze, where students and retirees are experimenting with agents that handle shopping, travel booking, and real-world tasks autonomously. Beijing already restricts OpenClaw use at state agencies over security concerns, but at least 4 municipalities are offering subsidies, making China the world's most contested agentic AI deployment battleground right now. Read more: Bloomberg
Meta Avocado AI slips past May: Meta pushed its next flagship model, codenamed Avocado, out of this month's launch window after internal benchmarks reportedly fell short of rivals like Google. The delay is notable given the billions Meta has poured into AI infrastructure and its recent hire of Scale AI's Alexandr Wang to overhaul the effort. Avocado will be Meta's first major model release under Wang's watch, so a performance gap at this stage signals the rebuild is taking longer than the company hoped. Read more: The Verge

Other things going on in the world of AI:

Yang wants AI firms taxed, not workers: Andrew Yang tells CNBC the US should stop taxing labor and redirect that burden to automation beneficiaries as AI threatens 50% of entry-level white-collar jobs. Read more: Business Insider

Tesla bucks AI layoff trend, grows headcount: While Atlassian cuts 10% and Block axes 4,000 citing AI, Musk says Tesla will increase staff as Optimus robots push output-per-human "nutty high." Read more: Business Insider

AI co-scientists accelerate climate research: Google DeepMind's Gemini co-authored 42% of an AMOC study, cutting review time 10x, while Google's new Groundsource tool predicts flash floods with 82% accuracy. Read more: Bloomberg

SaaS firms axe staff to fix SBC bloat: Falling software valuations and AI-driven revenue slowdown force companies like Atlassian (10% cuts) to slash headcount and reduce stock-based compensation before shareholder dilution bites. Read more: Business Insider

// EOF

That's today's build. Same time tomorrow — slightly out of step with the hype, right where the signal lives.

If this was useful, forward it to someone who ships things. If it wasn't, reply and tell us why — we're counting from zero either way.