I did talk about NOx and SOx in my "why gas turbines are bad" section, and guess it kinda goes hand in hand with the carbon capture where their waste is pure CO2.
Land and water use was something that I missed here and got happily educated about by K.R. himself during the earnings call.
Combined heat and power I actually didn't know about at all. Very interesting.
Man I kick myself for not looking at bloom quicker. Saw the name floating around, thought it was a hype pre-commercial kind of company until I dug slightly deeper and saw how incredible the tech actually is.
Fast set up, leap-frogs regulatory blockages, clean, scalable. Well done for the guys who got in. Right now valuations look like they're pricing in perfect but I wouldn't be surprised if the company keeps performing, they seem like the only player in the space able to qualm the bottleneck, and that means they have pricing power.
Caterpillar said the energy bottleneck is likely to only get worse going into 2030. Maybe its worth nibbling on any small dip.
Re Bloom - how concerned are you regarding Bloom’s need for chromium in their fuel cells. I read that by the time they ramp to 7GW, they ll be using a quarter of the refined chromium the world produces. Given the slow pace of growing rare metals refining - does this supply constraint bother you when discussing a potentially exponential growth story? From what I ve seen there isn’t an obvious alternative to chromium
On May 20–21 in London, Data Centre LIVE will bring together the companies, engineers, operators and infrastructure leaders shaping the next industrial layer of the digital economy: hyperscale data centers, energy systems, cooling technologies, semiconductors, optical networks and AI infrastructure.
Ceres Power, MHI; FCell, Topsoe, will all try to have a piece of this pie. Its hard for me to imagine BE keeping 100% of the market share for much longer.
Also on your comparision against Renewables I think you missed the case Renewables+ Bateries. Check the case of google in Pine Island just with renewables+ bateries ( Form Energy). The cost of bateries is falling quickly with and incredible learning curve. I Guess that is the way because we can not power all this GWs with fossil fuels. The case for carbon capture will just not happen any time soon.
The data center power thesis on Bloom makes sense until you stress-test the cost curve. Fuel cells win on reliability and grid independence. But the capital cost per megawatt is still the ceiling relative to natural gas at scale. Watching whether hyperscaler procurement teams start writing dedicated Bloom contracts is the real tell.
The “capital vs permission” framing is the key insight here. AI infrastructure has moved from asking who can afford to build to who can actually get power deployed fast enough. Bloom looks expensive only if time, permitting, conversion loss, and reliability are treated as footnotes instead of the whole game.
Yes, that's a massive part of the thesis. Read my earnings review. Bloom is the only scalable BTM energy company. Low capex like semicaps. Add capacity easily. KR tells us this every qtr.
Also a skeptic of Bloom being clean claims that Bloom officially told the world that its energy servers emit 884 pounds of carbon dioxide per megawatt hour.” A brand-new combined-cycle power station from Siemens, GE or Mitsubishi emits only 730 pounds of CO2 per megawatt-hour.”
I want to love Bloom, but I have a skeptic and I have a question. The question: I did read with new orders they have gotten from hyperscalers they are pretty much at max capacity in terms of production of new boxes for a while . Is there a way they are going to be able to fill more orders? In other words are they capable expanding capacity to meet this challenge or is this wrong?
You missed a few important other features of Bloom 😛-
1. Emissions. Order of magnitude lower NOx and SOx. Big deal for air permits.
2. Land use and curb appeal. Quiet and unobtrusive and use very little space if stacked up. Important for urban sites.
3. CHP. Combined heat and power. Many applications need heat not cooling like DCs. Can reach 80-90% efficiency if your use case needs it.
4. Water consumption. Matters in places like the New Mexico Oracle Project Jupiter.
Like I've said elsewhere it truly is the SSD or the LED bulb of power production. It's a bit easier to invert and say what it's worse at.
Mr. Bloom himself!
I did talk about NOx and SOx in my "why gas turbines are bad" section, and guess it kinda goes hand in hand with the carbon capture where their waste is pure CO2.
Land and water use was something that I missed here and got happily educated about by K.R. himself during the earnings call.
Combined heat and power I actually didn't know about at all. Very interesting.
Just a bit of good humored needling :) I did miss that you talked about it in the inference section as well.
Man I kick myself for not looking at bloom quicker. Saw the name floating around, thought it was a hype pre-commercial kind of company until I dug slightly deeper and saw how incredible the tech actually is.
Fast set up, leap-frogs regulatory blockages, clean, scalable. Well done for the guys who got in. Right now valuations look like they're pricing in perfect but I wouldn't be surprised if the company keeps performing, they seem like the only player in the space able to qualm the bottleneck, and that means they have pricing power.
Caterpillar said the energy bottleneck is likely to only get worse going into 2030. Maybe its worth nibbling on any small dip.
Jason, thanks for all the great work.
Re Bloom - how concerned are you regarding Bloom’s need for chromium in their fuel cells. I read that by the time they ramp to 7GW, they ll be using a quarter of the refined chromium the world produces. Given the slow pace of growing rare metals refining - does this supply constraint bother you when discussing a potentially exponential growth story? From what I ve seen there isn’t an obvious alternative to chromium
On May 20–21 in London, Data Centre LIVE will bring together the companies, engineers, operators and infrastructure leaders shaping the next industrial layer of the digital economy: hyperscale data centers, energy systems, cooling technologies, semiconductors, optical networks and AI infrastructure.
https://substack.com/@risksignalai/note/c-257164630?r=82bznj&utm_source=notes-share-action&utm_medium=web
Ceres Power, MHI; FCell, Topsoe, will all try to have a piece of this pie. Its hard for me to imagine BE keeping 100% of the market share for much longer.
Also on your comparision against Renewables I think you missed the case Renewables+ Bateries. Check the case of google in Pine Island just with renewables+ bateries ( Form Energy). The cost of bateries is falling quickly with and incredible learning curve. I Guess that is the way because we can not power all this GWs with fossil fuels. The case for carbon capture will just not happen any time soon.
I love Bloom Energy also. Thank you for putting this out.
The data center power thesis on Bloom makes sense until you stress-test the cost curve. Fuel cells win on reliability and grid independence. But the capital cost per megawatt is still the ceiling relative to natural gas at scale. Watching whether hyperscaler procurement teams start writing dedicated Bloom contracts is the real tell.
The “capital vs permission” framing is the key insight here. AI infrastructure has moved from asking who can afford to build to who can actually get power deployed fast enough. Bloom looks expensive only if time, permitting, conversion loss, and reliability are treated as footnotes instead of the whole game.
Can they address the order backlog? If not growth at this pace would be unsustainable
Yes, that's a massive part of the thesis. Read my earnings review. Bloom is the only scalable BTM energy company. Low capex like semicaps. Add capacity easily. KR tells us this every qtr.
Here is a detailed link to the Bloom hater. I know you are very astute so briefly what does this guy get wrong
https://melifinance.substack.com/p/bloom-energy-inc-this-time-is-different?r=em9cv&utm_medium=ios
Read whole thing
"This time is not different" is wrong.
This time is in fact different
That's about it
Also a skeptic of Bloom being clean claims that Bloom officially told the world that its energy servers emit 884 pounds of carbon dioxide per megawatt hour.” A brand-new combined-cycle power station from Siemens, GE or Mitsubishi emits only 730 pounds of CO2 per megawatt-hour.”
I want to love Bloom, but I have a skeptic and I have a question. The question: I did read with new orders they have gotten from hyperscalers they are pretty much at max capacity in terms of production of new boxes for a while . Is there a way they are going to be able to fill more orders? In other words are they capable expanding capacity to meet this challenge or is this wrong?
Adding hundreds of mw per qtr and scaling 10x without adding employees as they said on earnings
They will be fine
As scalable as the semi caps
it is hard to believe that Bloom does not have a real fuel cell competitor till 2030.
Any thoughts on FCEL
No one close to Bloom rn
They have 20 year tech lead in fuel cells