Is expected data center demand for natural gas going to cause a tighter gas market?
Likely, given the pace of growth
There are already about 2,000 data centers in the United States.
The emergence of AI as a driving force for Big Tech is accelerating demand for data centers, with some estimates suggesting a 40 - 50% growth rate for the next few years. With grid capacity under stress already and many communities resisting data centers in their areas owing to the risk of brownouts and blackouts, one likely outcome is rising demand for natural gas fired electricity generation, even co-generations, to ensure large data centers have a reliable and uninterruptible supply of power, which they need since they run 7 x 24.
According to an article in “Natural Gas Intelligence” by Carolyn Davis published on the NGI site on May 30, 2024, the 2,000 data centers in operation now consume about 20 Gigawatts of electric power.
Each 7 Gigawatts of power demand is equivalent to about 1 Bcf/day of natural gas demand if the power is all natural-gas fired. Davis puts the estimated capital costs to build a projected 70 Gigawatts of data center power at about $700 billion. Davis is using data from Carson Kearl, an energy transition analyst, with the that capital outlay expected to be in place by 2028. A single hyperscaler data center with a 200 Megawatt capacity would have a capital cost of 200/70000 x 700,000 = $2 billion if those broad numbers are right.
Natural gas demand to supply that 200 Megawatt facility would be 200/7000 x 1 Bcf/day = 29 Mmcf/day which, at $6.00 a Gigajoule, amounts to 29,000 x 6 = $174,000 a day or about $63 million per year.
Wholesale electric power runs from about $0.10 to $0.30 per KwH in United States. A 200 Megawatt facility running 7 x 24 consumes 200,000 x 365 x 24 = 1,752,000,000 KwH per year which from the grid at wholesale prices amounts to between $175.2 million and $525.6 million. Clearly, the savings from natural gas at $6.00 a gigajoule are material (in the range of $110 million to $450 million a year). I am sure data center operators would rather pay $63 million for power than pay three to eight times that much.
Most Alberta natural gas producers would be more than pleased to supply natural gas at a fixed price of $6 a gigajoule on a long term contract. At that price, for example, Peyto Exploration & Development (PEY.TO) would have annual cash flow of about CDN$1.5 billion and rival Birchcliff energy (BIR.TO) about CDN$850 million - if they received $6.00 a gigajoule for all their natural gas output.
The $6.00 a gigajoule price is arbitrary, chosen by me for convenience, since it is high enough to make WCSB producers very profitable and low enough to make a compelling case to Big Tech operators of data centers. There are savings in transportation (which I have ignored) if the data centers were located in the oil patch, since it is virtually free to transport data over fiber optics but costs about $1 to $1.50 a gigajoule to ship natural gas by pipeline to the U.S.A. For a single 200 Megawatt data center, the transportation trade off likely adds another $10 to $15 million to annual savings.
TC Energy sees opportunity to supply data centers with gas along its pipeline route, according to recent reports. But the real opportunity for both the data center operator and for producers in the Western Canadian Sedimentary Basin is a direct deal between them and data center operators without the need to ship the natural gas anywhere except to a local data center built on land in the oil patch through a dedicated pipe.
Will it happen? Under Trudeau, likely not since he consistently does whatever he can to punish Alberta for existing, but if Poilievre become Prime Minister in 2025, then I can see this unfolding quite quickly. Alberta Premier Danielle Smith sees the opportunity, and it is time for Canada to seize it. With at least some estimates putting the scale of opportunity at $100 billion, it is big enough to matter not only to Alberta but also to Canada.
If only we had a Prime Minister with enough common sense to drop the pretense that CO2 causes climate change, embrace the energy industry and seize the opportunities that would make Canada an economic powerhouse rather than an economic basket case. Maybe we will after the 2025 election runs Trudeau out of office?
mike - thanks for your excellent energy recommendations. can you value Pfizer (PFE) and Cameco (CCJ)? I bought cameco in Oct 2022, and am up 50 % (should have sold when it doubled). I bought it based on recommendations from Rick Rule whose advice has been extremely valuable. Nonetheless, I now think its not easy to value that company, and I am inclined to sell. Having worked out the metrics of Pfizer (thanks to Ashwath D's little book) I think it is worth 40-50 dollars per share. The thesis is not.unlike BCE's. It is as cheap as it gets, pays a nice dividend, sound balance sheet, has invested its COVID windfall (although that's hard to assess), and the company won't go away anytime soon. What I don't like is the ten year history of the company.
I am new to oil, but find the valuation exercises much easier than in tech (I work in machine learning/self driving cars). Going over the valuation exercises from a few places (Shubham Garg is one of them).
I've been doing online Co pilot research work on data centers and impact on potential players for the coming energy crisis stemmed from the AI data centers bubble and the potential impact on existing natty and oil positions along + infrastructure positions BIP.UN ( over $7B data centers under construction and commitment ) , + TRP natty and CPX gas fired co gen power plants
FYI research notes form co-pilot 4.0
USA Data Centers: 2,913 data centers across the United States spread across various states.
As of 2024, Canada boasts a total of 336 data centers and total global 5500 Data centers
A+