Technology
Data centers · Micron · China tariffs
5 min read7 min listen
Listen to the podcast
US voters approved data-center bans or restrictions in 40 of 41 local ballot measures across 10 states, while Micron halted construction of Taiwan's Fab 11E on November 5. Together with Texas governor-elect Gina Hinojosa's pledged approvals moratorium, these developments suggest that permission to build and labor agreements are becoming constraints on future AI capacity. Fab 11E had been scheduled to start production in 2028, so its pause does not establish a current memory-output loss. Meanwhile, USTR had published no extension as of November 8 for 178 tariff exclusions expiring late November 9; without one, covered entries from November 10 face restored duties, primarily 25% or 7.5%.
Approval constraints suggest financing alone will not secure AI capacity delivery: voters approved restrictions in 40 of 41 local measures, and Texas's governor-elect pledged an approvals moratorium.
USTR had published no extension as of November 8 for 178 exclusions expiring at 11:59 p.m. EDT on November 9. Without an extension, covered consumption entries or warehouse withdrawals from November 10 revert to Section 301 duties, primarily 25% or 7.5%.
Browse a few recent editions, or open the full archive.
Texas's proposed ratepayer indemnities and mandatory union labor suggest that developers could face a higher cost base even when projects receive approval.
Fab 11E's planned 2028 commercial start indicates that Micron's construction pause threatens future DRAM and HBM capacity, rather than demonstrating a loss of current production.
AI deployment depends on complementary facilities becoming usable together. A delayed data center can leave equipment and customer commitments waiting on power and permits; delayed memory expansion can leave a completed facility without its expected component supply. Spending on one part of that chain therefore cannot necessarily compensate for a bottleneck elsewhere.
Unresolved variables that could shift the assessment materially.
Decisions this bears on
The underweighted distinction is between financing-driven retrenchment and projects deferred by permission or bargaining constraints. The evidence does not establish weaker AI demand or quantify financing-led capital-expenditure cuts. Treating every interrupted project as the same kind of investment slowdown obscures that distinction.
4 sources
2 sources
2 sources
2 sources
Indicators and developments to monitor in the coming days.
T1 Primary · T2 Analytic · T3 Commentary