Physix Frontier · News Briefing Card (enbrief · Sep 20, 2026)

AI Talent Market Splits: Premium for Top and Edge, Middle Layer Depreciates

KEY FACTS

Yao Xingcheng left Tencent to join Thinking Machines Lab with a salary exceeding his previous tens of millions yuan annual package.
OpenAI is hiring Forward Deployed Engineers in Singapore, prioritizing production adoption over model benchmarks.
MCP protocol SDK monthly downloads approach 500 million as tool connectivity rapidly standardizes.
Codex user growth among non-technical departments surged 189-fold, signaling deep enterprise AI integration.

KEY DATA

Tens of millions RMBYao Xingcheng's former Tencent annual salary

Nearly 500 millionMCP SDK monthly downloads

189xCodex non-developer user growth

PHYSIX OBSERVATION

The value of AI talent is undergoing structural reshaping. Scarcity no longer stems from 'knowing how to use tools' but from solving extreme complexity issues that cannot be standardized (such as foundational training) or bridging the last mile (such as enterprise deployment). The middle layer loses its premium due to framework maturity; practitioners must beware of skill homogenization, as migrating to high-barrier domains at both ends is key to survival.

Source: enbrief original report ↗