Institutional long only equity investor. mostly cyclicals. Former Sellside ER. Strong dislike of pods.
Joined January 2023
- Tweets1.8K
- Following1.4K
- Followers823
- Likes471
LongOnly retweeted
Big development for $SOLS + $CC (also $INV $VRT $ETN):
Nvidia Director of Data Center Cooling and Infrastructure Ali Heydari showcased OMNICOOL at the ARPA-E Energy Innovation Summit, a 250kW pod cooled by pumped two-phase R-515B and R-1234yf (supplied by $SOLS + $CC) achieving PUE <1.05 and ZERO water consumption. Heydari is on the record that PG25 has limits and this validates two-phase on the $NVDA roadmap (note this project started 4 years ago).
This was done in collaboration with $VRT and $ETN (Boyd). $VRT is also a $INV partner. Incrementally positive for these names but is another datapoint that this is a competitive market and $INV / Zutacore likely will not be alone in two-phase cooling. IMO refrigerant players Solstice Advanced Materials $SOLS and Chemours $CC are the best way to play two-phase cooling, as they are positioned to win regardless of vendor.
LongOnly retweeted
Twitter all day: Omg, he's crazy, he's gonna start WWIII
Twitter now: Lol he TACO'd again!
LongOnly retweeted
$SOLS should be a winner here. Can't see the US doing this but relying on the Canadian and French for their nuclear fuel supply chain.
This times 1 billion
My hot take is that financial modeling will never be fully automated by AI
The best analysts already know that all models are wrong. At best, good models are just directionally correct.
Despite this, the reason good analysts still build models from scratch is because it fundamentally teaches you what drives the business you are looking at
To build a good financial model, you have to understand:
> what is unit economics of the business?
> what is price and volume mix?
> what decisions will the management make to grow top line?
> what segments are growing or declining, and why?
> how will margins trend and why?
> what type of working capital does the business need?
> what capital expenditures will be needed in the future?
> how will the growth be funded?
And much more.
As you go and build your model, you have to think through each of these aspects of the business is great detail.
Not just the historical data, but rather how each one trends over the next 3-5 years in the future
This is ultimately what gives you a fundamental understanding of the business that you cannot learn by simply reading a 10K or an equity research report
In fact, equity research analysts already publish their own models that you can steal if you have access to them
Most hedge funds do have access, and yet, hedge fund analysts still build their own models internally from scratch
Similarly, if you outsource building your model to an AI, you will still be severely limited in your conceptual depth of knowledge on the company
Sure, you might use AI to tweak models here and there, or make tedious edits to an excel file
But my hunch is that the best analysts will continue to build financial models till the end of time
That will never truly go away
1
1
642
LongOnly retweeted
My hot take is that financial modeling will never be fully automated by AI
The best analysts already know that all models are wrong. At best, good models are just directionally correct.
Despite this, the reason good analysts still build models from scratch is because it fundamentally teaches you what drives the business you are looking at
To build a good financial model, you have to understand:
> what is unit economics of the business?
> what is price and volume mix?
> what decisions will the management make to grow top line?
> what segments are growing or declining, and why?
> how will margins trend and why?
> what type of working capital does the business need?
> what capital expenditures will be needed in the future?
> how will the growth be funded?
And much more.
As you go and build your model, you have to think through each of these aspects of the business is great detail.
Not just the historical data, but rather how each one trends over the next 3-5 years in the future
This is ultimately what gives you a fundamental understanding of the business that you cannot learn by simply reading a 10K or an equity research report
In fact, equity research analysts already publish their own models that you can steal if you have access to them
Most hedge funds do have access, and yet, hedge fund analysts still build their own models internally from scratch
Similarly, if you outsource building your model to an AI, you will still be severely limited in your conceptual depth of knowledge on the company
Sure, you might use AI to tweak models here and there, or make tedious edits to an excel file
But my hunch is that the best analysts will continue to build financial models till the end of time
That will never truly go away
$SOLS rallying as the government is looking for hosts for “nuclear lifecycle innovation campuses”.
$SOLS is still too cheap. Article here.
energy.gov/articles/departme…
This, a million times.
Also, I can just download any sellside model I want. But if you don’t know how it (or the business) works, it’s utterly useless
This quoted post is unavailable.
5
563
LongOnly retweeted
$SOLS keeps grinding higher.
Not because it builds data centers.
Because it supplies what modern data centers are increasingly constrained by.
(1/8)
LongOnly retweeted
Solstice Advanced Materials $SOLS shared good info re their UF6 conversion business at the Morgan Stanley Chemicals Conference and Goldman Sachs Industrials Conferences:
- Annual UF6 capacity is 8,000 tons, going to 10,000 tons in 2026 w room to go to up 15,000 tons.
- Backlog is $2.2B, majority of which signed in 2023-2024 after restarting the Metropolis Works facility.
- 90% of the business is secured through LTAs with escalators linked to uranium spot prices and inflation. 10% of capacity is kept for the spot market.
- Margins are within 200-300 bps of the segment total and are comparable to those at $CCJ Cameco Fuel Services.
Am long $SOLS. Refrigerants is an underappreciated space benefiting from rate cuts when they come, the AI data center buildout, and regulatory changes. Also owns the only UF6 conversion site in the US, which is positive positioning for nuclear + strategic investment.
LongOnly retweeted
More thoughts re $SOLS:
At 10,000 tons, $SOLS is 16% of global UF6 capacity. This rises to 35% when excluding Russia and China, with Orano SA and Cameco $CCJ providing 26,000 tons of capacity. This makes $SOLS Metropolis Works UF6 conversion facility a critical national asset in the nuclear supply chain.
Solstice Advanced Materials $SOLS shared good info re their UF6 conversion business at the Morgan Stanley Chemicals Conference and Goldman Sachs Industrials Conferences:
- Annual UF6 capacity is 8,000 tons, going to 10,000 tons in 2026 w room to go to up 15,000 tons.
- Backlog is $2.2B, majority of which signed in 2023-2024 after restarting the Metropolis Works facility.
- 90% of the business is secured through LTAs with escalators linked to uranium spot prices and inflation. 10% of capacity is kept for the spot market.
- Margins are within 200-300 bps of the segment total and are comparable to those at $CCJ Cameco Fuel Services.
LongOnly retweeted
If he throws a TD I’ll give away a signed Giants Jersey to someone that likes and retweets this.
Go @Giants
LongOnly retweeted
This was preventable.
The #Giants actions made it inevitable