Water water everywhere, Nor any drop to drink
May 24, 2026 – weekly comment
***************************
I cry water, water, water
The truth to shame the liar – Eric Burdon, Water
In March of 2000, the based-on-a-true-story movie Erin Brockovich was released. Julia Roberts famously played the lead role. Erin was an unemployed mother of three. She was injured in a car crash and filed a lawsuit with lawyer Ed Masry. She loses the suit, but ends up going to work for Masry, who hires her even though she has no legal experience.
Julia Roberts and Erin Brockovich
From Wikipedia:
Erin is given files for a real estate case where the Pacific Gas and Electric Comany (PG&E) is offering to purchase the home of Donna Jensen, a Hinkley resident. Erin is surprised to see medical records in the file and visits Donna, who explains that she had kept all her PG&E correspondence together. Donna appreciates PG&E’s help: she has had several tumors, and her husband has Hodgkin’s lymphoma, but PG&E has always supplied a doctor at their own expense. Erin asks why they would do that, and Donna replies, “Because of the chromium”. Erin begins digging into the case and finds evidence that the groundwater in Hinkley is seriously contaminated with carcinogenic hexavalent chromium, but PG&E has been telling Hinkley residents that they use a safer form of chromium. After sev’l days away from the office doing this research, she is fired by Ed until he realizes she has been working the entire time and sees what she has found, and rehires her.
The case resulted in the largest direct-action lawsuit award in history at that time against PG&E, $333 million, in 1996. Heck of a movie!
What made me think of this movie, and of the Flint Michigan water crisis which occurred in 2014 and resulted in a $626 million verdict, was this posting on X:
From @HedgieMarkets
https://x.com/HedgieMarkets/status/2057901307596312725
Brown water is coming out of kitchen taps in Morgan County, Georgia, and residents say it started when explosive blasting began at a nearby data center construction site. EPA Assistant Administrator for Water Jessica Kramer agreed under oath at a congressional hearing this week to investigate.
Morgan County is now the seventh community I have written about this month dealing with measurable harm from data center construction. Saline Township, Tucson, Fayetteville, Lake Tahoe, Phoenix, Newnan in Coweta County, and now Morgan County all share the same story. The hyperscalers picked locations where land was cheap, regulations were light, and local governments were eager for the tax revenue.
Then there’s this weekend article on BBG:
Hyperscaler Debt Flood Brings Derivatives Bonanza
https://blinks.bloomberg.com/news/stories/TEXUWTT96OSL
As big tech companies raise hundreds of billions to fund artificial intelligence investments, Wall St banks are increasingly finding they have to trade more credit derivatives to keep doing business with the hyperscalers.
The banks’ purchases of CDS protection allows them to increase business with hyperscalers. “Investors continue to look to buy hyperscaler CDS, and a lot of that does come from CVA desks” (Credit Valuation Adjustment desks)
CDS tied to MSFT, AMZN and ORCL notched $4.6b in notional trading volume in the first quarter, from $759 million a year earler… META CDS- only launched last October – had $534m in notional trades, more than double the prior quarter.
These numbers don’t seem particularly large or worrying to me. But the water issue could loom large given the unpopularity of data centers. As WH Auden wrote:
“Thousands have lived without love, not one without water.”
From Google (Gemini)
Grid Impact: Nationally, U.S. data centers withdraw roughly 0.3% to 0.4% of total daily U.S. water withdrawals.
Household Equivalents: A medium-sized data center (around 10–20 MW) uses about 110 million gallons annually, matching the daily usage of roughly 1,000 average American households.[this make ZERO sense to me even though it was just cut and paste from AI! The average American household uses about 300 gallons per day (seems high) so the sentence SHOULD read,”… matching the ANNUAL usage of roughly 1000 households”]
Again, the numbers don’t seem alarming. But when combined with other snippets of data like a new historic low in University of Michigan Consumer Sentiment at 44.2, and polls on opposition to data centers (see below), the prospect of an army of Erin Brockoviches seem more likely.
Public opposition to data centers has escalated into a widespread national movement, with 70% of Americans opposing the construction of artificial intelligence (AI) data centers in their local communities according to Gallup polling data.
Pope Leo XIV’s first encyclical, Magnifica Humanitas, is set for release tomorrow, Monday, May 25, 2026. It will address “the protection of the human person in the age of artificial intelligence”.
Look, it never pays to bet against the Pope.
With the tech industry at an overall >50% weighting in the SP500, and the fact that hyperscaler AI investment is expected at $2.59 trillion in 2026 (Gartner), a significant retrenchment of AI stocks will reverberate throughout the economy. The CDS numbers aren’t big, but what IF some of the hedge funds have to pay out on hyperscaler defaults? Counterparty risk?
Finally, there’s this:
DeepSeek To Make Permanent 75% Discount on Flagship AI Model (BBG)
I don’t have any particular trades to suggest, though a large reversal in tech could easily flip sentiment in rates to a theme of disinflation. Currently 0QZ6 (midcurve option on Dec’27) 9675/9725 cs settled 4.75 ref Z7 9604.5. On March 2, as Iran hostilities started, SFRZ7 traded as high as 9709.5. I’d be inclined to pick up some cheap call spreads on red SFOR contracts on any further cheapening of the contracts.
| 5/15/2026 | 5/22/2026 | chg | ||
| UST 2Y | 408.1 | 412.7 | 4.6 | wi 413.0 |
| UST 5Y | 425.8 | 426.7 | 0.9 | wi 427.4 |
| UST 10Y | 459.5 | 457.0 | -2.5 | |
| UST 30Y | 512.8 | 508.1 | -4.7 | |
| GERM 2Y | 273.5 | 264.1 | -9.4 | |
| GERM 10Y | 316.6 | 303.7 | -12.9 | |
| JPN 20Y | 365.0 | 368.5 | 3.5 | |
| CHINA 10Y | 175.6 | 174.8 | -0.8 | |
| SOFR M6/M7 | 35.0 | 37.8 | 2.8 | |
| SOFR M7/M8 | -10.5 | -15.5 | -5.0 | |
| SOFR M8/M9 | 5.0 | 4.5 | -0.5 | |
| EUR | 116.21 | 116.03 | -0.18 | |
| CRUDE (CLN6) | 101.02 | 96.60 | -4.42 | |
| SPX | 7408.50 | 7473.47 | 64.97 | 0.9% |
| VIX | 18.43 | 16.70 | -1.73 | |
| MOVE | 79.87 | 78.43 | -1.44 | |
