Descriptions:
The All-In podcast hosts break down three Anthropic announcements in February 2026 that triggered cascading losses across the software sector. On February 3rd, Anthropic launched a legal plugin for Claude integrated with Thomson Reuters, LexisNexis, and LegalZoom — sending all three stocks down more than 10%. On February 20th, a limited research preview of Claude Code Security caused CrowdStrike, Okta, and related cybersecurity names to sell off sharply. Then on February 23rd, Anthropic announced Claude’s ability to modernize legacy COBOL codebases — the 60-year-old language that runs 95% of U.S. ATMs, processes Social Security payments, and underpins most of global banking infrastructure — sending IBM to its worst single trading day since 2000 and erasing $31 billion in market capitalization.
Chamath Palihapitiya and the panel debate whether this represents tactical hedge fund deleveraging or a structural repricing of enterprise software companies whose cash flows AI is beginning to credibly threaten. The discussion turns to a framework for how public markets price “when” risk — the point at which previously reliable revenue streams become uncertain — and how AI is accelerating that uncertainty across legal tech, cybersecurity, and legacy infrastructure simultaneously.
The episode also covers the broader economics of token pricing, with predictions of a 90% reduction in output token costs by year-end, the emergence of a new role — “AI operations” managers who can train and manage agents — and why infrastructure constraints on land, power, and chips may moderate the pace of disruption even as model capabilities advance rapidly.
📺 Source: All-In Podcast · Published February 28, 2026
🏷️ Format: News Analysis
