Professor of Mathematical Finance at UC Santa Barbara. CFMAR Co-director. Working on renewable energy, computational finance & ML in Actuarial Science.
California, USA
Joined December 2019
- Tweets278
- Following430
- Followers230
- Likes145
Mike Ludkovski retweeted
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
dwarkesh.com/p/openai-huggin…
Citi vs Citadel: who is going to be right about the Fed int rate decision this week⁉️ Bloomberg headlines today:
A new day and a new record for SPP load: now over 60.4GW and trending up. Real time prices surging as we speak.
#SPP is about to crack the record of 60GW of load. 4 of the highest-ever load days have been in the past week!
#SPP is about to crack the record of 60GW of load. 4 of the highest-ever load days have been in the past week!
MISO DA prices are above $800 for this evening. Interesting to see how RT performs. After all the winter price spikes driven by cold, we are now in the summer peak load season.
Got an email that @UCSB campus is under an Electrical Demand Reduction Event today 7/16 with CAISO issuing notice of potential generation capacity shortage due to high regional demand. UCSB is asked to reduce power 4-7PM. Yet CAISO forecasts peak load <41GW, nothing extreme...
Thanks to SunZia wind project in New Mexico, the solar-dominated #CAISO is enjoying a major uptick in wind generation. All ten of top-10 wind days have occurred in the past month! @grid_status
I will be doing a deeper dive into CAISO wind soon.
Two milestones for CAISO in the past two days. On Friday afternoon, a new record of 30.5GW was set for max renewables generation. On Saturday at 1pm, a new record low net load of -9.1GW was reached. Both are enabled by batteries 🔋 which now routinely contribute +/-10GW of power.
The day of minimum net loads: all three of CAISO (West), NYISO (New York) and PJM (Mid Atlantic+Midwest) set record low net loads today. In CAISO, thanks to the batteries the grid can go almost -8GW nowadays.
Crazy congestion in CAISO right now along Moss Landing-Metcalf Path 15. Prices are $-60 in Monterey Bay area and above $400 in Gilroy and Morgan Hill, leveling off as you approach San Jose. @grid_status
MISO set a new renewable generation record yesterday, over 39GW. This is a leap of 2GW+ compared to old record set 5 weeks ago.
For comparison, ERCOT's renewables high is 47GW set in June 2025 and CAISO's high is 29GW also from June 2025. MISO might catch ERCOT next year...
Just three days after the previous record, SPP solar gen broke the 2GW barrier on April 3. The new all-time solar generation record of 2085 MW is 400MW higher, equivalent to a 17.1% jump. I am sure multiple further records are forthcoming in the next few weeks.
SPP is finally adding solar generators. There have been no fewer than 19‼️new max solar generation records set in SPP in 2026. With the days getting longer, more records are definitely forthcoming. With peak solar gen of <2GW, SPP is still very far behind neighboring MISO & ERCOT
CAISO has surpassed 12 GW of discharging this past Sunday. At that peak discharging (~7pm), batteries were providing nearly 44% of total power. See the near-perfect complementarity between solar falling off & storage picking it up below, erasing the vast majority of evening ramp.
A tell-tale sign of increasing share of renewable generation are new records of *minimum* net load. Today it was MISO whose all-time lowest net load decreased by 2+ GW. Last week it was ERCOT, where net load trough is approaching zero. All top-10 lows were set in the past month.
ALT Dates of minimum net load in ERCOT per gridstatus.io
It's been windy in the High Plains, which is very good for renewable generation in SPP. Wind has supplied 58%-70% of load throughout the past 36 hours, quite exceptional consistency (note how little solar there is -- it was practically nonexistent in SPP until last year)
