The debate over using HTML versus Markdown to communicate with AI agents reveals a deeper shift. The primary job of a knowledge worker is no longer to complete a task, but to create the optimal conditions, context, and scaffolding for an AI agent to perform the work effectively.
HTML excels in the new "agent staging" paradigm because its native features (tabs, color-coding, expandable sections) can encode a project's "mixed doneness." This visually distinguishes between locked requirements, open exploratory areas, and provisional decisions, calibrating the agent's autonomy more effectively than plain text.
Chipmaker TSMC's recent sales growth, while still high, was half of what analysts expected. This isn't a sign of weakening AI demand. Instead, it indicates that TSMC has hit its physical manufacturing capacity limits and cannot keep up with the frenetic pace of orders, a bullish signal for the industry.
In a radical attempt to address the drastic AI compute shortage, major housing developers like PulteGroup are testing the installation of micro data centers on newly built homes. These units would function as nodes in a distributed computing cluster, highlighting that every possible avenue is being explored for more compute power.
Once TSMC's top customer, Apple has signed a chip-making deal with Intel, partly due to White House pressure but also because the AI boom has consumed TSMC's capacity. This move illustrates that extreme demand for AI chips is diminishing the negotiating power of even the world's largest tech companies.
OpenAI competitor Anthropic is seeing massive investor demand for its next funding round, partly because its recent deal with SpaceX is viewed as having "dramatically de-risked" the investment. Some investors believe this partnership has resolved Anthropic's most significant bottleneck by securing its access to compute power.