Nov 8, 2026 · Research
How often does a power date come true?
New large-load sites in Texas met 13 of 59 first power dates. Half waited 221 days or more past the date for power.
Read the brief
More research
-
How often does a power date come true?
New large-load sites in Texas met 13 of 59 first power dates. Half waited 221 days or more past the date for power.
-
PJM filed its backstop cost split with each zone’s share fixed for 15 years
Utilities serving new data centers and their customers carry the cost. PJM asked for comments by November 30 and a FERC order by January 29.
-
PJM’s backstop could cost $1.4 billion a year. Zones with new data centers pay, at a share fixed for 15 years.
PJM files the new cost split with federal regulators by October 29. Each utility’s share is set from the large loads it reports a week before bidding opens, and then it does not move.
-
Texas energization holds and proposed transmission billing
ERCOT’s August report identified 17 projects awaiting energization approval. A separate proposal would begin transmission billing when capacity becomes available.
-
PJM’s capacity shortfall and FERC’s September and October orders
FERC accepted and suspended the backstop procurement, rejected interim service, and retained the large load registry.
-
SPP’s federal review and the Texas connection
SPP’s large load rules face federal review. NERC’s separate registry proposal is now at FERC.
-
Documentation and financial security under §25.194
The adopted rule permits forfeiture when supporting documents cannot verify information. The remedy is discretionary; audit duties and eligibility paths matter.
-
Computing costs and the limits of grid access
A conditional scenario about computing costs, heat removal and grid access, with the June 2026 ERCOT queue as a historical example.
-
Reading the ERCOT large load queue
ERCOT’s June 2026 snapshot records 466,497 MW of requests and a 5,700 MW Observed Energized tier. Their 1.2% ratio is not a construction completion rate.
Lab notes
Working notes from our agent-systems experiments. Not energy research.
-
Automation in investment research
An archival note on automated research, portfolio decisions and human oversight.
-
MCP and the interface between agents and tools
An archival technical note on tool access, protocol design and integration boundaries.
-
A2A and communication between agents
An archival technical note on agent discovery, task exchange and coordination.
-
The systems around an AI agent
An archival technical note on orchestration, state, tool execution and oversight.