China curbs AI talent travel + SpaceX IPO and orbital data centers

Reuters, citing Bloomberg News, reports that Beijing has widened informal travel restrictions originally placed on senior DeepSeek researchers to AI talent at Alibaba and other private firms — requiring some professionals to seek official approval before traveling abroad, with the policy framed around state-secret concerns and strategically important AI work. The move treats AI researchers themselves as restricted assets, a national-security frame that mirrors the logic the US has used in reverse for chip export controls. The pattern hardens a two-way decoupling at the human-capital layer rather than just the supply chain.

In this week's Stratechery analysis, Ben Thompson argues that SpaceX's rumored IPO at a $2 trillion valuation only makes sense if Starship enables data centers in orbit. His core thesis: terrestrial data center expansion is now constrained more by community zoning opposition than by power generation, and the existing Starlink V2 Mini satellite form factor — about 7.4 meters by 2.7 meters — is comparable to NVIDIA's NVL72 rack. Combined with Starlink's laser interconnects, the constellation already has the network topology required for distributed orbital compute; power dissipation and radiation hardening become engineering problems rather than fundamental obstacles to agentic-inference workloads.

DuckDuckGo's US app installs rose 18.1 percent week-over-week between May 13 and May 25, peaking at a 30.5 percent jump on May 25 — with iOS installs averaging 33 percent growth and peaking near 70 percent — after Google replaced its blue-link results with an AI agent that answers queries, executes tasks, and runs in the background. Visits to the company's AI-free subdomain grew about 22.7 percent week-over-week. CEO Gabriel Weinberg said Google is force-feeding AI with no opt-out. DuckDuckGo holds about 2 percent of US search; the absolute numbers are small, but the trajectory is the first measurable consumer reaction to the Google AI Mode rollout.

An ICLR 2026 paper from Amazon Web Services introduces two complementary techniques — Set-Supervised Fine-Tuning and Global Forking Policy Optimization — that train language models to generate multiple distinct reasoning paths for the same problem rather than collapsing onto a single strategy. On the American Invitational Mathematics Examination 2025 benchmark, the approach reaches roughly 64 percent accuracy, a 6.84-point gain over the standard supervised-fine-tuning-plus-reinforcement-learning baseline; comparable gains land on the AIME 2024 set and the LiveCodeBench coding benchmark. The result chips away at a long-standing critique that verifiable-reward reinforcement learning converges on narrow solution strategies and loses the diversity that lets multi-attempt sampling outperform single-shot.

Anthropic named KiYoung Choi — a 30-year technology executive previously running Snowflake's Korea operation, with country leadership stints at Google Cloud, Adobe, Autodesk, and Microsoft — to lead its incoming Seoul office, with senior leadership traveling to Korea in the coming weeks to officially open the team. The local operation will work with Korean enterprises, government and research institutions, and the developer community. Anthropic disclosed that Korean users engage with Claude at over three-and-a-half-times the rate expected for the population size; Law and Company and SK Telecom are named production deployments.

OpenRouter, the multi-model API gateway founded in 2023, raised a $113 million Series B led by CapitalG — Google's growth venture fund — at a $1.3 billion post-money valuation, more than doubling from its $547 million Series A last June. The company routes requests across more than 400 models from Anthropic, Google, OpenAI, xAI, and DeepSeek, and now serves roughly 8 million users at around 100 trillion tokens per month — a 5-times jump in throughput over the past six months. The funding signals that multi-model routing has moved from optional plumbing to default architecture for production AI deployments.

Human Archive, founded by Berkeley and Stanford students and backed by Y Combinator, raised $8.2 million from Wing Venture Capital, NVP Capital, Y Combinator, and angels from OpenAI, NVIDIA, and Google. The startup equips home-services, hotel, and restaurant workers across India with camera-enabled caps and motion-capture sensors to collect first-person task data for training robots — more than 1,000 active headsets and over 50 custom sensor rigs are deployed today, with customers receiving discounted service rates in exchange for consent. Workers are paid roughly one dollar per hour, and several major Indian gig platforms — Urban Company and Pronto among them — declined to participate. The economics raise familiar questions about who captures value when human labor becomes physical-AI training data.