Pinned Tweet
Non ascoltate me, ascoltate @repubblica - forse sarebbe ora che nPlan si attivasse anche in Italia per prevenire altri disastri? Per non parlare del ponte sullo stretto...
repubblica.it/tecnologia/202…
People ask how I did a PhD in deep learning theory while working as a quant.
Badly. Slowly. Every weekend, for years.
'Brick by brick' isn't a motivational poster. It's a calendar you don't skip.
1
2
79
Well done to @JensenHuang for literally screwing $200b away. Hell of an escort pricing 😂
26
Maybe the reason we should actually celebrate Steve Jobs as a great founder is that he appears to be the only one that didn't cheat on his wife?
17
OpenAI now calls its data centres "Compute Campuses".
Fitting. Lambda says its $33bn Alabama campus takes up to five years to build.
In Italy, students who finish late are "fuori corso". 77% of data centre projects are already behind plan at their first update. Fuori corso.
15
Sam Altman says your PC's RAM idles at 5% and belongs in his data centre.
By that logic my golf clubs, idle for the 20 years I rage-quit, belong at Francesco Molinari's house.
I need to work on my short game more.
36
Disney: find 20% savings.
Also Disney: ~$9bn a year of capex, inside a ~$60bn parks plan.
The real money isn't in headcount. Shave 1% of overrun off that plan and you've found $600m. 🧵
1
48
We rebuilt the schedule for Piston Peak, the Cars land at Magic Kingdom, from permits and aerial photos only.
Our plan: 14 Mar 2030
Forecast P50: 24 Feb 2031, 347 days late
Chance of on time: 1%
Half the slip lands after the mountain is done: ride testing and certification.
1
36
Delay gets baked in years before anyone gets a layoff letter. Fix the critical path, not the org chart.
Full study: nplan.io/disney-cars-land
Independent, public data only. No involvement or endorsement by Disney.
44
Alan Mosca retweeted
Founder work-life balance is just forgetting you have a family while making sure everyone knows how hard you work.
The @TheB1M did a video about the Hard Rock guitar hotel in Vegas. Completely forgot to mention that they already built one in FL before they even owned the land from the Mirage. And it worked so well in FL that they decided to hire the same architects and reproduce it but bigger
1
2
67
OPG announced Darlington Unit 1 as 142 days early. It finished 44 days after its working plan.
Both are true. OPG keeps three dates and announces against the public one, which held 4-7 months of buffer.
Fine on a job you've done four times. Not on a first-of-a-kind SMR.
1
34
We forecast the new Darlington SMR from public filings only.
Public date: end 2030, ~9 months of buffer.
Most likely date we see: August 2032.
Six decisions can win back ~9 months. Four are due before 1 Dec 2026.
If you report progress, ask which date you're early against.
1
18
Free 17-page study and plan file: nplan.io/darlington-smr
Fully independent, public data only. No involvement from or endorsement by OPG or its partners.
17
Alan Mosca retweeted
Halloween in Australia 😂😂
Readers added context they thought people might want to know
This video is AI-generated, not real footage of Halloween in Australia.
leadstories.com/hoax-alert/202…
kangarooislandtoursaustralia.com.au/blog/is-it-tru…
Alan Mosca retweeted
What in the fuck are we doing
77% of data center projects are already behind plan at their first update.
Not at the end. At the first update. Mostly permits and long-lead equipment.
The delay is baked in before anyone pours concrete. 🧵
2
1
58
How we know: at @nPlanHQ we looked at 211,620 finished pieces of data center work, each with its planned and its actual duration.
'Behind plan' = work done trails time used by 10+ points.
Same data: 1 in 4 construction tasks takes at least twice as long as planned.
10