Black Hills

IThe business

Black Hills Corporation delivers natural gas and electricity. It is headquartered in Rapid City, South Dakota, and was incorporated there in 1941.³ The company serves about 227,000 electric customers in Colorado, Montana, South Dakota and Wyoming, and about 1,138,000 gas customers in Arkansas, Colorado, Iowa, Kansas, Nebraska and Wyoming.³ That is 1.37 million customers in eight states, run by 2,795 employees, a quarter of them in unions.⁶³ It trades on the New York Stock Exchange as BKH. If the merger closes, the parent is renamed Bright Horizon Energy Corporation and trades as HZN on the same exchange.⁶

The model is simple to say and hard to copy. A state commission lets the company charge rates that cover its costs and a fair return on the money it has invested, called the rate base. The company then invests more, asks for higher rates, and the return grows with the rate base. The costs of fuel and purchased gas pass through to customers with no mark-up.³ So revenue mostly follows commodity prices and is not what matters. What matters is the rate base, the return allowed on it and how fast the commission lets it grow.

In 2025 the company made $2.31B of revenue, $537.5M of operating income and $291.6M of profit for common shareholders, or $3.98 a share. Adjusted for $0.12 of merger costs it earned $4.10.⁴¹⁴ Through nine months of 2026 revenue was $1.70B, operating income $397.1M and adjusted earnings $2.89 a share, against $2.68 a year earlier. Management reaffirmed its 2026 guidance of $4.25 to $4.45 on 4 November.²

Gas is the larger half

Gas Utilities brought in $1.38B of 2025 revenue and $320.8M of operating income. Electric Utilities brought in $943M and $222.5M.³ So gas is 59% of operating income C (calculated). That surprises people who think of Black Hills as a Wyoming power story. The gas side owns 4,581 miles of transmission pipe, 44,840 miles of mains and service lines, seven storage sites and 494 miles of gathering lines. The electric side owns 1,386 MW of generation and 9,478 miles of lines.³

Gas also carries the larger share of the capital plan. Of $4.67B planned for 2026 to 2030, gas is $2.50B and electric $2.04B C (calculated).³ Most of the gas money goes to replacing old pipe. The company says it has removed all cast and wrought iron from its systems and keeps replacing aging pipe under programs that regulators let it recover between rate cases.³

Where the growth is

The growth that gets attention is electric load in Wyoming. Wyoming Electric set four peak records this year. The latest, 439 MW on 20 July 2026, was 16% above the 2025 peak.² In the third quarter the electric utilities sold 20% more industrial power than a year earlier C (calculated), which the release attributes mainly to Wyoming’s large-load customers.² In February management said its pipeline of data center requests had tripled during 2025 to more than three gigawatts, with Meta ramping up and Microsoft growing, together about 600 MW to be served by 2030.⁹

On 30 September it signed definitive agreements to serve a Google data center in Cheyenne. The company will invest $1.8B in 564 MW of new gas generation, between 2027 and 2029, and manage the output of about 2.1 GW of third-party resources. It expects about $150M of net income from the project in 2030, and $2.4B of free cash flow after the $1.8B of investment, through 2048.¹⁰ Google has already advanced $399M, which the company expects to repay by 30 June 2027.¹⁰ I value this in Section VI. What matters here is that it sits in a non-regulated affiliate, with one customer, under a contract.¹⁰

What changed in the last month

Three things. On 6 October the Google contract was announced, and the stock rose from $70.74 to $75.84 the next day.¹⁰¹ On 22 October the companies named the merged company and said that every approval except Montana’s was in.⁶ And after the close on 4 November the company reported its third quarter, with adjusted earnings of $0.57 a share against $0.45, helped by $0.15 of new rates and $0.06 of favourable weather. It also received the South Dakota approval of a rate settlement the same day.² The stock had closed that day at $84.19, up 2.4%, before the report, so that rise was not a reaction to it.¹

The NorthWestern merger

The merger has not closed. It is pending, and the one condition left is a vote of the Montana Public Service Commission. The companies expect to close by the end of 2026.²⁶ The companies announced it on 19 August 2025 as an all-stock merger of equals. Each NorthWestern share becomes 0.98 Black Hills shares, and Black Hills shareholders end up with about 56% of the combined company, NorthWestern holders 44%.⁸ NorthWestern had 61,517,850 shares at 24 July 2026, so about 60.3 million new Black Hills shares would be issued, on top of 76.25 million today C (calculated).¹⁵⁵

Here is what it does to the share and the price. First, the share count rises 79%, from 76.3 million to about 136.5 million C (calculated).⁵¹⁵ A Black Hills holder keeps every share. But the share is now a smaller piece of a bigger company: 56% of the whole, where before it was 100% of Black Hills. Second, the price. At today’s closes, 0.98 of a Black Hills share is worth $82.51 and a NorthWestern share trades at $82.00, a gap of 0.6% C (calculated).¹¹⁸ So the market prices the two stocks almost exactly at the exchange ratio, which fits a high chance of closing. It does not tell me what either stock would be worth alone. The $84.19 therefore holds three things I cannot separate: the Black Hills business, the Google news, and a deal the market expects to close. Third, the numbers per share. NorthWestern guides to $3.68 to $3.83 of adjusted earnings and pays $2.68 a year.¹⁶ Black Hills guides to $4.25 to $4.45 and pays $2.812.² Adding the earnings, I get about $4.13 a share for the combined company before any savings, against $4.35 for Black Hills alone C (calculated). The companies say it will be accretive in the first year.⁸ I can only reproduce that if the savings are kept, and Montana’s draft order puts them at $48M to $70M.¹³ I come back to this in Section VI.

IIThe moat

The protection here is a legal one. Black Hills holds exclusive franchises to deliver gas and electricity in its territories, and a new competitor cannot lay pipe next to it. That is strong, in the sense of how hard it is to compete away. It gives no pricing power. The same commissions that protect the franchise set the price, and the company cannot raise rates without asking.³ So the question is not whether Black Hills keeps its customers. It is whether the return it is allowed stays above what investors need.

What regulators allow

The 10-K lists the authorized return on equity for each unit. They run from 9.3% to 9.9% where they are stated, with equity of 46% to 53% of the capital structure in the state cases.³ Three jurisdictions use black-box settlements, which fix a total revenue number and do not state a return. South Dakota Electric, Iowa Gas and Kansas Gas are on that basis.³ That hides the return, so I cannot check it.

The regulators have not given the company what it asked. In Arkansas the company asked for $29.4M of new revenue and a 10.5% return, and received $17.9M and 9.8%, 61% of the dollars and 0.7 points less on the return.⁹² In Wyoming, South Dakota Electric got $4.0M at 9.60%. In South Dakota it got $27.4M on a 7.58% weighted cost of capital, with a four-year pause in base rates until August 2030.² The pause is the part I would watch: it fixes revenue while costs rise. Colorado Electric has asked for $26.7M at 10.5% on a 51% equity share, with a decision expected in the first quarter of 2027.² The Arkansas outcome says Colorado will probably be lower than 10.5% E (estimate).

Earned against allowed

In 2025 Black Hills earned $291.6M on average common equity of $3.66B, or 8.0% C (calculated).⁴ That is below every allowed return in the table. One reason is plain. Goodwill of $1.30B, 34% of year-end equity, comes mainly from the 2016 purchase of SourceGas, and regulators do not let the company earn a return on it.³⁴¹⁹ So the equity that earns is smaller than the equity on the balance sheet. The combined release says the merger doubles each company’s rate base to a total of $11.4B, which puts Black Hills near $5.7B E (estimate).⁸ Half of that is $2.85B of equity, and 9.6% on it is $274M, close to the $292M the company earned C (calculated). That ties out, which is why I use $5.7B and half equity.

Evidence it is holding

The record is of a business that gets most of what it asks for, a little late. Operating income grew every year since 2021, from $409M to $538M.⁴ New rates and rider recovery added $0.95 a share in 2025 and $0.63 in the first nine months of 2026.¹⁴² The dividend has risen for 56 consecutive years, the second-longest streak in the industry, the company says.² Wyoming passed wildfire liability protection in March 2025, the 10-K cites Montana too, and South Dakota followed in early 2026. Each protects a utility that follows an approved wildfire plan.³² I found nothing equivalent for Colorado in the filings, and Colorado Electric has the most electric customers, 102,152 at year end.³

Evidence it is narrowing

Three things. First, affordability. Montana’s commission heard nearly 30 members of the public ask it to reject the merger, citing data centers and bills.⁷ Affordability is the live issue in the one state still to rule. Second, earnings have not kept pace with investment. Adjusted EPS rose from $3.74 in 2021 to $4.10 in 2025, 2.3% a year, while capital spending ran well above depreciation.¹¹⁴ Third, the legacy cost of money is running out. Three notes of $400M each, due in 2027, 2029 and 2030, carry coupons of 3.15%, 3.05% and 2.50%.³ Replaced with debt at 6%, they would cost about $37M more a year in interest, or $0.42 a share after tax E (estimate). That is 10% of 2026 earnings, and regulators will pay it back through rates only with a lag.

IIIThe competitive advantage

Utilities do not compete with each other in a territory. They compete for capital, and they compete for permission. On those two things Black Hills has some advantages and some weaknesses, and I try to be plain about which are which.

Advantage one: spread across regulators

The company reports in eight states and many commissions. No single commission can make or break the company. The combined firm would be even more spread: the announcement says no one jurisdiction would be more than 33% of the business.⁸ That is real protection. It also means the company has to win many separate rate cases in different political settings, and a bad one in a small state costs little.

Advantage two: a data center tariff that avoids the usual trap

The usual risk of building for a data center is stranded cost: the plant is built and the customer leaves. Wyoming Electric offers large customers a tariff, called Large Power Contract Service, under which the customer carries the cost. For the Google project, Mr. Evans says the arrangements are structured so that Google bears all costs of serving the data center for the life of the project.¹⁰ The company also holds $399M of Google’s money in advance.¹⁰ This is a good design. It also means the return is a contract return, not a regulated one, and the project is held in a non-regulated affiliate.¹⁰ If Google or its partners fail, the protection is the contract, not the commission.

Advantage three: the cost of the next project

Because Black Hills already has the Cheyenne Prairie generating site and a 260-mile transmission line between South Dakota and Wyoming called Ready Wyoming, built in three years, it can move faster than a company starting from nothing.⁹¹⁰ Management says its tariff provides speed to market. That matters when buyers of power are racing to be connected.⁹

Where it is weaker

Scale. Black Hills has about $5.7B of rate base E (estimate) and a market value of $6.4B. The merged firm would have about $11.4B of rate base and about $11.5B of market value at today’s prices C (calculated).⁸ Larger utilities can often fund a $2.5B data center build with less new equity as a share of the total. Black Hills is a price taker for capital: its notes due 2034 and 2035 carry 6.15% and 6.00%, and the five-year note it sold in October 2025 carries 4.55%.³

This table shows why I credit NorthWestern’s own pipeline before I say Black Hills holders lose in the swap. NorthWestern has signed development agreements for 150 MW of data center load from late 2027, with a possible 1,500 MW or more, and it has asked Montana’s commission for a large-new-load rule, which is pending.¹⁶ That is a real option, and it is not in the guidance. My bull case lets NorthWestern grow at the top of its range, 6%, and my base uses 5%.

The other regulated utilities on this site

None of them is a close match. Enbridge, a pipeline and utility company, has a price that asks 9.4% a year on my cash flows against the 9.0% I require.²⁰ Black Hills asks 7.6% against 8.25%. The two required rates differ because the risks differ, and I do not claim that one is better value. The point is only that a reader who wants a dividend-paying infrastructure name has other places to look on this site.

IVManagement

Linn Evans was president and chief executive in every year from 2021 to 2025, and Kimberly Nooney is chief financial officer.¹¹ They run the company through the merger, and then they step aside. The agreement says Brian Bird, NorthWestern’s chief executive, becomes chief executive of the combined company, with Marne Jones, now Black Hills’ chief utility officer, as chief operating officer, NorthWestern’s Crystal Lail as chief financial officer, and Ms. Nooney as chief integration officer.¹² Mr. Evans stays through the close and then leaves.¹¹ So the people I am judging are mostly the people who will not run the company.

How they are paid, and what that pays them to do

The annual bonus rests 70% on adjusted earnings per share. The rest is safety, reliability and customer measures.¹¹ For 2025 earnings of $4.10 paid out at 107% of target, and the whole bonus at 118.4% of target.¹¹ The adjusted figure leaves out merger costs, so the team is not penalized for the deal.¹¹ Longer-term pay is restricted stock and performance units. For 2025 to 2027 the units depend 40% on how the stock does against a group of peers, 30% on average adjusted EPS, 20% on non-fuel operating cost per customer and 10% on emissions cuts.¹¹

What does that pay them to do? To grow earnings per share, and in a utility the way to grow earnings is to grow the rate base. Every dollar invested earns an allowed return, so a bonus on EPS rewards investing more, as long as the commission allows it. The cost-per-customer measure is the check. It pays them to restrain operating cost, though not the amount invested. I do not say that is wrong. It is how a regulated company grows. It means the pay plan is neutral on whether a given project is a good use of shareholders’ money, and the commissions have to be the check.

The share-price test has been poor. For 2023 to 2025 the stock ranked at the 20th percentile of peers, and that part of the award paid nothing. The whole performance award paid 29.97% of target.¹¹ The weight on that test was then cut from 70% to 40% for the 2025 to 2027 plan.¹¹ The 2025 to 2027 plan was set before the 2023 to 2025 results were known, so I do not read motive into it.

Mr. Evans’ 2025 pay was $5.99M against a median employee at $129,748, a ratio of 46 to 1.¹¹

How much of their own money is in

Mr. Evans owns 174,272 shares, worth $14.7M at $84.19, including 36,578 restricted shares C (calculated).¹¹¹ The company requires its chief executive to own six times salary. At a $1.0M salary that is $6M, so he owns about 2.4 times the rule.¹¹ Ms. Nooney and Ms. Jones own about 38,700 shares each, $3.3M each C (calculated). All directors and officers together own 427,823 shares, including 95,370 of equity awards, less than 1% of the company.¹¹ The two largest holders are index managers, BlackRock at 14.6% and Vanguard at 12.0% by their last filings, which are dated 2024 and 2025.¹¹ That is real skin, though not enough to dominate how he thinks. The more important part of his wealth is what happens at closing: his package on a closing, valued at $60.16 a share in the proxy, was $11.5M, of which $8.4M is vesting equity.¹¹ At $84.19 the equity is worth more E (estimate). Performance units from before August 2025 convert at the greater of target or actual.¹¹ So the deal pays the team to complete it. I do not think that is corrupt. I do think it means the board, not management, has to decide whether the exchange ratio was right.

What they did with capital over a full cycle

The largest decision of the decade was the 2016 purchase of SourceGas, closed on 12 February for about $1.89B including $760M of assumed debt.¹⁹ Goodwill rose from $360M to $1.30B that year, the company raised $1.77B of new long-term debt, and total assets went from $4.6B to $6.5B.⁴ The filings I opened do not split out what the purchase contributes today. It also left $1.30B of goodwill that regulators do not let the company earn on.

Since then capital has gone into regulated assets at regulated returns, which is a safe use. The cost is in the per-share figures. From 2017 to 2025 diluted shares rose 33%, earnings per share rose 24% and the dividend per share rose 49% C (calculated).⁴ So the payout went from 55% of earnings to 68% C (calculated). Over ten years the company sold $1.36B of stock and paid $1.40B of dividends.⁴ A holder who took the dividend in cash was, in effect, paid with other holders’ new money. That is how a growing utility works, and it is why growth in the share price has been slow: a share bought in early November 2016 for $60.26 is worth $84.19 now, plus about $23 of dividends, or about 6% a year without reinvesting C (calculated).¹⁴²

The pattern is steady, not poor. Book equity per share rose faster than earnings per share, which is another way of saying that return on equity slipped. Management’s stated aim is 4% to 6% growth in EPS from a 2023 base of $3.75, with the 2026 guidance at 6% over 2025.¹⁴ If it delivers the middle of 2026 guidance, EPS will be 5.1% a year above that base C (calculated). From the $3.93 it earned in 2023 it is 3.4% a year.¹¹

VFinancials

Ten years of reported figures. Revenue includes the gas and power that customers pay for and the company passes through, so I lean on operating income and earnings.

What an owner could really take out

This is the number I care about most, and it is the hardest to pin. Reported profit in 2025 was $291.6M. The cash an owner could take out is what is left after the money needed just to keep the system running, and after stock pay of $11.8M.⁴ The company gives no figure for maintenance against growth spending, so I show a range. If replacing worn assets costs the same as depreciation, $283.8M, then 2025 owner cash was $377.8M, 30% above profit. If replacing at today’s prices costs 1.5 times depreciation, it was $235.9M, 19% below E (estimate).⁴ The truth is probably between those, and the range includes reported profit. What the company paid out, $197.9M, was covered either way.⁴

But there is a second way to read the cash. The company spent $819.8M in 2025 against $673.4M from operations, and paid $197.9M in dividends. The $344M shortfall, plus $219.2M of share sales, paid for the rest.⁴ Over ten years the gap after dividends was $2.48B. An owner who wants cash today gets the dividend, 3.3% of the price. An owner who wants growth lets the company keep reinvesting at about 9.6% on regulated equity, and funds part of it with fresh shares. Both are ways of saying the same thing: the free cash flow is negative, and the return comes from the rate base growing.

Quality of earnings

Four items I would not take at face value. Weather: mild weather cost $0.12 a share over nine months, even after a $0.06 gain in the third quarter.² Tax: the company guides to an effective rate of about 14% for 2026, below the 21% federal rate, mostly from wind tax credits.² At 21% the 2025 pretax income of $343.5M would have paid $72M of tax instead of $43.7M, a drop of $0.39 a share C (calculated).⁴ Regulatory assets: $378M at 30 June, of which the Winter Storm Uri balance was $50.7M at year end and shrinking.⁵³ Goodwill: $1.30B, 34% of equity, unchanged since 2016, and the pro forma balance sheet adds $1.44B more on a preliminary estimate.³²²

The earnings are real cash. The Uri episode shows how: in 2021 operating cash flow was −$64.6M, the company funded the gas and the commissions let it recover over time.⁴³ It is a reminder that earnings here depend on regulators repaying what a storm costs.

The balance sheet

At 30 June 2026, the latest balance sheet, because the third-quarter filing is not out, long-term debt was $3,993.9M, current maturities $410.0M and notes payable $90.0M, a total of $4,493.9M against $71.5M of cash.⁵ Debt over capital was 52.8% by my arithmetic, and the credit agreement’s version was 0.53 against a limit of 0.65.⁵ S&P rates the company BBB+ and Moody’s Baa2, both stable.⁵ Management targets funds from operations of 14% to 15% of debt, 100 basis points above its downgrade threshold of 13%.⁹ The filings do not give the exact figure.

Here is how leverage works on the shareholder. Net debt is $4.4B against a market value of equity of $6.42B, so debt is 40% of enterprise value C (calculated). A utility is meant to run like this, because stable earnings can carry debt. But the cost of that debt resets. About $1.55B comes due in 2027 to 2030, and I estimated above that refinancing the three low-coupon notes at 6% costs $0.42 a share. And $4.2B of the $4.735B of long-term debt sits at the parent, not at the utilities.³ The parent is paid from the dividends the utilities send up, which is why the rating agencies watch the parent’s cash flow.

The data center work adds more. The company says the $1.8B Google generation and $686M of Wyoming transmission come on top of the $4.7B plan, $2.5B in all.² It plans to fund the generation with project cash flow, debt and other financing, and has not said how much equity.¹⁰ In 2025 it sold $219M of stock, and for 2026 it guided to $50M to $70M, and sold $69M through the third quarter.⁴² If it funds half of $2.5B with stock at today’s price, that is about 15 million shares, or 19% more than today E (estimate). I do not know the plan. I know that it exists and that it moves the answer, which is why I value the contract as cash and not as earnings.

VIValuation

First, how this goes badly wrong

Before any number, I ask what would most likely ruin this as an investment. For a utility the dangers are few and they are not about competition. There are three.

The return stops following the investment. The company invests about $900M a year and earns an allowed return only if commissions agree the spending was prudent and let rates move. If they pull back, rate base grows and earnings do not. Earned return on book equity was already 8.0% in 2025 against 9.3% to 9.9% allowed.⁴³ Each half point of return on $2.85B of regulated equity is about $14M, or $0.19 a share, 4% of earnings E (estimate). A run of Colorado-style cuts, a South Dakota-style freeze until 2030, or a ruling that shifts data center costs to households would do it.²

A large uninsured loss. The 10-K says insurance may not cover wildfire, gas explosion, cyber or storm losses, or may not pay in time.³ Wyoming, Montana and South Dakota have liability protections for utilities that follow approved wildfire plans. I found none for Colorado, where the company has 102,152 electric customers.³ I cannot size this. One event could cost more than a year of earnings, and the odds in any year are small.

Paying too much for growth and funding it badly. The merger, the Google build and $4.7B of base investment all need new capital. $1.55B of debt due in 2027 to 2030 rolls from low coupons to about 6%, a cost of about $0.42 a share after tax E (estimate). If equity is sold at a lower price than today’s, existing holders carry that. And the price itself is a risk: a 19.4 times multiple on a utility, with a dividend yield 1.6 points below the 10-year Treasury.¹²³ If rates stay near 5% and the multiple settles at the 17.4 median of the last nine year-ends, the price would be about $76, with no change in earnings E (estimate).⁴¹

All three together are my bear case: earned return slips, growth runs at 3% and I require 9%. That gives $45, below the October 2023 low close of $47.93.¹

What the price implies

Now the reverse question. What has to be true for $84.19? I solve for each piece alone, holding the others at my base.

The price asks for about 8.1% a year in earnings growth for five years, or $6.42 of EPS in 2031. The company's own plan is 4% to 6%. So the price is not a plan price. It assumes the plan plus a good deal more, which means Google delivered in full and at least some of the 3 GW pipeline turning into contracts like it. Management points the same way: its latest release says large-load opportunities bring significant upside to the current five-year plan.² The price has risen $13.45 since 6 October, which at 76.25 million shares is $1.03B, and $389M of that came on 7 October, the day after the announcement.¹⁵ I value the whole Google contract at $766M at 7.5%, or $536M at the 70% weight in my base E (estimate). The market has paid more than that.

What I can’t judge

Several inputs matter a great deal, and I cannot know them. I list them so the range is not mistaken for precision.

• Montana’s decision. The commission is all Republican, staff recommend approval, a 3 to 2 vote moved the decision to 10 November so that Commissioner Fielder could draft a condition on reserving Colstrip power for Montana customers, and the legal deadline is 12 November.¹³⁷ I cannot tell what conditions will be attached or whether they will change the value. The share price already says approval is very likely, with a 0.6% spread.¹¹⁸
• Colorado. The company asked for 10.5%. It was allowed 9.3% to 9.5% last time.³² I do not know what the commission will do in the first quarter of 2027.
• Wyoming. The new transmission tariff, called the Large Customer Transmission Cost Adjustment Mechanism, decides who pays for $686M of lines. A decision is expected in the first quarter of 2027, and the two transmission projects in the middle of 2027.² I cannot judge how the commission will weigh households against data centers.
• Black-box settlements. South Dakota Electric, Iowa Gas and Kansas Gas settle on a revenue number with no return stated.³ I cannot check what return they provide.
• Politics. Public opposition to data centers, in Montana and elsewhere, is outside my model. The draft Montana order says costs will be addressed in separate cases.¹³
• Google and its partners. The contract is real and has a $399M advance behind it, but I cannot check the finances of the third-party resource owners the company will manage.¹⁰

The merger, in numbers

I value the shares as if the merger closes, since the companies and the market both expect it. I show the alternative too. To do this I value NorthWestern separately, with its own guidance: $3.755 of EPS in 2026, a 71% payout of $2.68, growth of 5.0% in my base, and the same discount rate. That gives $65 a NorthWestern share, or $66 per Black Hills share it converts into E (estimate). My Black Hills value, before the merger, is $77. So on my numbers a NorthWestern share turns into 0.98 Black Hills shares worth $76, against $65 for the NorthWestern share itself: its holders gain about $11 a share, and Black Hills holders pay for it. The ratio was fixed in August 2025, before Google, so about 44% of the Google value goes to NorthWestern holders for nothing, unless NorthWestern’s own data center agreements are worth as much.⁸¹⁶ I give those no value in the base.

The offset is savings. Montana’s draft order puts total savings at $48M to $70M, with $36M from labour, and I can’t see more because the company redacted the detail.¹³ Savings in a utility go to customers at the next rate case. I assume shareholders keep a part of them, with costs to get there, and value the net at +$100M in my base, −$150M in my bear and +$350M in my bull E (estimate). The merger costs Black Hills holders $4 a share in my base, and the combined company is worth $73 a share E (estimate).

The rate base lens

A second way to check is to value the equity that regulators allow the company to earn on. It is a cross-check, not the method.

It shows the gap in plain figures. If the company earned its allowed return on this equity for ever, growing at 3.5%, regulated equity would be worth about $3.7B, or $48 a share. The market pays $6.4B. My DDM says $77 for Black Hills alone. What fills the gap is the plan: rate base growing about 6% a year, for years, while commissions allow 9.6% against a cost of equity that I put at 8.25%. And the Google contract, which does not come from the regulated base at all. I cannot make this reconcile to the dollar, because rate base is my estimate. It is the reason I keep the range wide.

The growth in rate base is also my estimate. The plan invests $4.67B over five years. Depreciation, at the 2025 level of $284M growing 5% a year, takes about $1.65B. That adds about $3.0B to net plant of $8.23B, or 6.4% a year before deferred taxes.³⁵ NorthWestern says 4% to 6% rate base growth on a $3.2B plan.¹⁶ So I say near 6%, and the 5.5% I use for earnings is that less the dilution from new shares E (estimate).

The discount rate

I use 8.25% for the base. The basis is the 10-year Treasury at 4.95% plus 3.3 points.²³ The 3.3 is my judgement of the extra return an owner of a stable regulated utility needs over a government bond. Regulators allow 9.3% to 9.9%, which is 4.4 to 5.0 points over the Treasury, and I take a haircut because commissions usually lag. For the bear I use 9.0% and for the bull 7.5%. The bull rate is close to the price’s own implied rate on my base cash flows, 7.6%, so the bull assumes the market is right about the rate and I am wrong about growth. I use the same rate for the dividend stream and the terminal value, and I discount dividends from mid-year timing to today.

The grid is the honest statement. A one point move in the discount rate changes the value by roughly $12 to $19, about as much as the whole distance from base to the margin-of-safety price. Utilities trade on the rate more than on any forecast of earnings.

Three cases

The Google line needs a word. The company says $150M of net income in 2030, about 45% of 2026 earnings C (calculated).¹⁰² Taken at 19 times that would be a $2.9B gain, or $37 a share, and that is how it feels. But it costs $1.8B of capital and the plant is fully depreciated in 2048. The company puts the cash after the investment at $2.4B over the life.¹⁰ At 7.5% that is worth $10.04 a share. At 6.5% it is $11.90 and at 8.5% $8.39 E (estimate). The difference between earnings and value is the capital. I weight 70% in the base, because I cannot tell whether every approval comes in time, and 0 in the bear.

The cases are not equal bets. I weight them 25%, 50% and 25%, giving $73.

Other uses of the same dollar

The Treasury is the honest benchmark. The base case gives 7.6%, which is 2.6 points over a bond. That is a thin premium for the risks above, and it is why I do not call this cheap. Enbridge’s price offers 0.4 points more than I ask and Black Hills’ offers 0.7 points less, though they are different businesses and I would not rank them on that line alone.

What past lows looked like

The stock fell 44%, from a record close of $85.53 in February 2020 to $47.93 in October 2023, over three and a half years.¹ In March 2020 it closed at $50.93.¹ My bear value of $45 is about what the 2023 low was. The last time it closed at or below $60, my margin-of-safety price, was 2 October 2025, 13 months ago.¹ And the stock has risen 40% since, on the merger, the Google contract and a falling rate.¹ If those fade, it can fall back, and has twice since 2020.

The margin of safety

My rule is to ask for a price at which the bear value is at least three-quarters of what I pay. Bear divided by 0.75 gives $60. Below $60, a bear outcome costs 25% or less, and the base case offers 9.3% a year. That is 29% under today’s close, so today’s price is 40% above it. That is not close.

Two different questions

How good the business is and how good the price is are separate. The business is stable and well regulated by the standards of the industry: a dividend raised for 56 years, eight small jurisdictions, a clear route to growth. The price is a different matter. The shares are within 2% of their ten-year high, the dividend yield is further below the Treasury than at any of the last ten year ends, and the price needs the plan and more.¹²³ So the quality is good, and the price is high for it. A share bought on 2 October 2025 at $59.76 is up 41%, about the same as the 40% gain over the whole ten years from $60.26, so almost all of the decade’s price gain came in 13 months C (calculated).¹ A good business bought above its value can still earn 7.6% a year, and over a bond that is a modest return for taking the risks. I can’t say the market is wrong. It sees the data center pipeline, and I am valuing only the first contract.

VIIThe case for the price

What most people miss

Three facts favour the company and get little attention. Google has already put money down: $399M of refundable advances for long-lead equipment, which the company expects to repay by 30 June 2027, and the arrangement is structured so that Google bears all the costs of serving it.¹⁰ Electric demand in the Cheyenne area has risen for 20 years in a row, and the peak of 439 MW on 20 July 2026 was 16% above 2025.² And the need to issue shares has shrunk: $219M of stock in 2025, and a guided $50M to $70M for 2026.⁴⁹

The bull case

The strongest version is that Black Hills is a steady utility that has just found a way to grow faster than its plan, and that a stock at a ten-year high can still be cheap if the second and third contracts follow the first. Management says it expects to deliver in the upper half of its 4% to 6% growth target, and describes the large-load work as significant upside to the current five-year plan. It will give a 2027 to 2031 plan with its year-end results.² Scotiabank raised its rating and its target, to $81 from $66, on 7 November 2025, calling Wyoming data center demand high-probability. It estimated about 35% of earnings upside from a 1.8 GW project, and said the shares had limited downside if the merger failed.²⁴ I opened the report on TipRanks, not the Scotiabank note.

The wider sell-side has followed. MarketBeat shows an average target of $83.25 across 8 analysts, from $77 to $97.²⁵ That is a roll-up of other people’s views, and I treat it as such. It also sits at today’s price, so the typical target offers no upside. I found no public view on Black Hills from any named investor beyond the analysts above. BlackRock and Vanguard are listed in the March 2026 proxy at 14.6% (as of March 2025) and 12.0% (as of December 2023), but they are index managers and I do not count them as a bull case.¹¹

Against reality

If the plan grows EPS at 6.5% through 2031, Google delivers in full, and the merger costs what I think it does, my value is $80. Add half of a second contract of Google’s size and it is about $83 E (estimate). So the price is possible, and what it asks for is two things beyond the plan. Today, one is partly in hand. Google is signed, with the Wyoming approvals for $686M of transmission still ahead. The second does not exist yet: there is a pipeline above 3 GW, under confidentiality, and no signed contract beyond Google.²⁹ I put rough odds on the four things the price needs: EPS growth near 6.5%, 40%; Google in full and on time, 75%; a second contract worth half as much again, 50%; and the merger closing with the savings kept, 85%. Multiplied, that is 13% E (estimate). They overlap, so perhaps 15%. That is not an argument that the price is wrong. It is a measure of how much has to go right.

What a 10x would take. Ten times $84.19 is $841.90, or $64.2B of market value on today’s shares and $115B on the merged share count. At 20 times earnings, which is already generous for a utility, that needs about $3.2B of net income, eleven times the $292M of 2025. At the 2025 net margin of 12.6%, that is about $25B of revenue, eleven times the $2.3B it makes now C (calculated).⁴ By 2036 it needs net income to grow 25% a year for ten years, from a plan of 4% to 6%. The price rose 40% in the last ten years.¹ A regulator sets the return on each dollar invested, so earnings can grow faster than 6% only by investing much faster, and the customers pay for it. The merger doubles the size and not the value per share. I do not know of a regulated utility that has grown tenfold from this size, and for Black Hills it is close to impossible.

Where I come down

What moved me is the Google contract: a tariff that makes the customer carry the cost and a $399M advance make a risky project a reasonable one, worth about $10 a share at full weight. What did not move me is the price. It already pays for the contract, and part of a second, and the 3.3% yield sits well under the Treasury. What would raise my values is a signed second contract on similar terms, a Colorado ruling at 9.5% or better, and a funding plan for the Google build with little new equity. What would lower them is a heavy Montana condition or a freeze in Wyoming.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• The merger. It could still be rejected or approved on terms that hurt. Montana is the last state, and the end date moves to 18 November and then to 18 February 2027 only if everything else is satisfied.¹⁷⁷ If it fails, I value a share at $77 in my base, which is $4 higher, and at $46 in my bear. The price of $84.19 would probably fall, since it now reflects a deal at 0.6% from parity with NorthWestern.¹¹⁸ Each party owes $100M in some cases if the deal ends.¹⁷
• Integration. Mr. Evans leaves at closing, and the combined firm would be run by NorthWestern’s chief executive; Ms. Jones and Ms. Nooney stay in new roles.¹² NorthWestern took over Avista’s and Puget’s shares of Colstrip Units 3 and 4, a coal plant in Montana, for $0 on 1 January 2026. It reports about $18.0M a year of added operating cost for the Avista share, not yet in customer rates, and about $30.0M for the Puget share, largely offset by a contract that runs only to late 2027.¹⁶ Black Hills holders would own 56% of that company’s results.
• Data centers and the household bill. The Wyoming tariff and the Montana rule both rest on the claim that data centers pay their own way. If a commission decides otherwise, or if a customer fails, the cost lands somewhere. The Google generation sits in a non-regulated affiliate, with no commission to turn to.¹⁰
• Cost inflation and supply. The 10-K flags tariffs, transformers and generation equipment as a risk to the plan.³ A cost overrun on $2.5B of data center projects would fall on the affiliate that holds the plant and on shareholders.
• Wildfire. Protection exists in three states and I found none for Colorado.³
• Weather. Mild weather cost $0.12 a share in the first nine months, even after a $0.06 gain in the third quarter. Weather evens out over time and I do not treat it as a trend C (calculated).²
• Interest rates. About $1.55B of debt reprices by 2030, and the multiple could fall if the 10-year Treasury stays near 5%.³²³
• Tax credits. A 14% tax rate is helped by wind credits; a change in the law would cost about $0.39 a share if the rate rose to 21% C (calculated).²⁴