ProPetro

IThe business

ProPetro is two businesses on one balance sheet, and they are at very different ages. The old one pumps. It provides hydraulic fracturing, wireline and cementing services to oil and gas producers, all of it in the Permian Basin of Texas and New Mexico, and it was 67% fracturing by revenue in the first nine months of 2026.² The new one, PROPWR, was formed in December 2024. It buys gas-fired engines and turbines, puts them on a customer's site and charges for the power. It earned $1.5M of revenue in all of 2025 and $22M in the third quarter of 2026 alone.⁵³

The company came to the market in March 2017, selling 13.25M shares for $170M after costs.⁴ It was a fast-growing pumper then, with a customer list led by Pioneer Natural Resources. In December 2018 it bought Pioneer's pressure-pumping assets for 16.6M shares and $110M in cash, and Pioneer became a large customer and its largest holder.⁴ Exxon Mobil bought Pioneer in May 2024 and sold all 16.6M of those shares at $16.66 on 20 May 2026.²¹¹ Exxon was still 24.9% of 2025 revenue, and the contract for two of the company's electric fleets with Exxon's XTO unit is expected to end in late 2026 without renewal.⁵²

The pumping business

At 30 September the company owned 1,205,500 hydraulic horsepower: 452,500 of dual-fuel units that burn diesel and gas, 435,000 of older diesel units, 6,000 of gas-only and 312,000 of electric FORCE units that it rents under leases.² It has 28 wireline units and 31 cementing units.² It runs 13 active fleets, up from 11 at the start of the year, and says it is essentially sold out; it estimates that only the mid-70s Permian fleets are active in the whole basin.⁸ More than two-thirds of its active horsepower is on contract, and most of those contracts renew in the first half of 2027.⁸ That date matters for the valuation: if pricing has risen, the repricing flows straight to profit. It is the company's best-supported source of upside, and the one I can least measure in advance.

Fracturing is a business of high capital and thin pricing power. Customers pick on price first, the 10-K says, then on reputation, emissions and service.⁵ The top five customers were 68.2% of 2025 revenue and the top ten 84.5%.⁵ In the third quarter, fracturing earned $45.4M of segment Adjusted EBITDA on $235.2M of revenue, a 19% margin; wireline earned 22% and cementing 19%; unallocated corporate cost was $13.7M.³ After corporate cost the completions side made 15.6% in the quarter, and the company says rebuilding capacity, rehiring crews and restocking parts cost more than it expected.³⁸

PROPWR

PROPWR sells "power as a service": it supplies and runs the engines and the customer buys the gas. Roughly 80% of the equipment it has on order is gas reciprocating engines and 20% modular turbines.² The first large customer was a hyperscaler data centre in the US Midwest, 60 MW with batteries, announced in October 2025 and running since June 2026.⁷¹⁶ Since then contracts have been signed with Coterra, an unnamed upstream operator, an industrial customer and, in September, a subsidiary of Targa Resources for about 230 MW to serve gas-processing plants in the Permian, fully deployed in early 2028.¹⁷¹⁶¹⁸

I looked for the number that most people skip. The company does not list the 510 MW by customer. The only data-centre contract in its releases is the 60 MW one, which I take to be inside the 510 MW, so 60 MW of 510 MW is a data centre, 12% of the total, and the same company says the majority of its fleet will go to data centres.⁷⁸ The remainder is oil and gas, midstream and industrial work that carries shorter terms; the January update put the weighted tenor at about five years.¹⁷ The October presentation targets 10 to 15 year terms for the larger plants it hopes to build.²⁰ Management says it is in "advanced negotiations" on several hundred megawatts of data-centre power and that timelines have lengthened.⁸ Those talks have run for at least two quarters without a signature, and I cannot tell from here when they end.

How it earns

The power segment earned $7.6M of Adjusted EBITDA on $22.3M of revenue in the third quarter, a 34% margin, after a $0.7M loss in the second.³ Management targets a payback of four to six years on $1.4M to $1.5M a megawatt, which is $230k to $375k of EBITDA a megawatt a year, and says oil and gas and midstream work is nearer four.⁸ I do not yet know what the realised number is. The company has not said how many megawatts earned revenue in the quarter; it guides to 250 to 275 MW running at the end of 2026.³ Assuming about 175 MW on average, the third quarter ran at roughly $170k a megawatt a year, below target, which is normal for a first year but is the one data point that exists E (estimate).

How it is paid for

The company says it raised about $1.5B for PROPWR in two years.⁸ In January 2026 it sold 17.25M shares at $10.00 for $163M net.²²¹ In May it sold $690M of 0.00% convertible notes due November 2031, convertible at about $23.17, with a capped call to $29.49 that cost $36.8M.²² It has an asset-based revolver of $350M, a Caterpillar equipment loan, and a $350M Stonebriar lease line that is so far undrawn.²²³ At 30 September it held $729M of cash against $862M of debt at face, and liquidity of $934M.³

IIThe moat

The test is what stops a rival from earning what ProPetro earns. I look in four places and find real protection in only one, and only partly.

Pumping. None. Price is the first factor customers use, the equipment can be bought, and the 10-K says the markets are highly competitive.⁵ Scale and a good crew help, and the shortage of high-quality horsepower after years of attrition helps more at the moment, but that is a cycle, not protection. Operating income was positive in five of the nine years from 2017 to 2025 C (calculated).⁶ A business with real protection does not lose money when the rig count falls; ProPetro did, on an operating basis, in 2020, 2021, 2022 and 2024.

Supply. This is the genuine one. Engine and turbine supply is tight, and in April 2026 the company signed a framework with Caterpillar to reserve about 1.5 GW of engines, with an option for about 600 MW more through 2031, against a minimum purchase of about $1.1B.²⁴ Of the roughly 2.6 GW it calls secured, more than 1 GW is Caterpillar's 10 MW engine.⁸ A rival without a slot cannot match the delivery date. But supply is a scarce input, not a customer lock, and it comes with an obligation, below.

Switching costs. Once an engine farm is running on a customer's site, moving it is expensive. That is true of every on-site power plant and it only protects a contract while it lasts. PROPWR's terms for early exit are not public, so I cannot say what a customer pays to leave.

Operating record. One year, about 250 MW running by the end of 2026, one hyperscaler site. The company is right that this matters: it is one of few behind-the-meter providers delivering prime power to a data centre today.⁸ It is not enough yet to count as protection. A competitor with more units, longer history and a lower cost of capital can rent the same engines.

What would show protection is a recontracting price that holds. The oil and gas contracts average about five years, so the first ones end from about 2030.¹⁷ Nothing in the filings tests that.

IIIThe competitive advantage

The advantage ProPetro claims is the combination: it already works in the Permian oil patch, it knows how to run engines in a field, and it can offer a customer that does not want to wait for the grid a plant in 12 to 18 months from a signed contract, against three to seven years or more for a utility connection.²⁰ I believe the speed claim. The hyperscaler site went from a signed contract in October 2025 to running load in June 2026.⁷¹⁶

The second advantage is the balance sheet, which is better than at most small oilfield companies. It has $729M of cash and no debt on its revolver.³ It also has a customer, Targa, that it calls investment grade and that sits in the same basin.¹⁸ That counterparty quality matters as much as the engine count, because the whole programme is financed on contracts.

The third is the cost side of the old business. The five electric fleets cost $63M a year in rent, which sits inside Adjusted EBITDA.³ The first buyout is expected to cost $15M to $20M in the fourth quarter, with at least three more in 2027; I assume the same price for each.³² Each buyout turns about $12.6M a year of rent into a one-off cheque of about $17M E (estimate). For a company with $192M of Adjusted EBITDA, five such buyouts are worth $63M of EBITDA, a third of the total. That is a quiet source of earnings, and I put it in the base case.

An honest list of disadvantages is longer. The company is small beside Caterpillar's other customers. Its access to capital depends on a share price that fell 58% from its 2026 closing high of $18.20 on 29 April to $7.73.¹ And it is signing megawatts into a market that other, bigger firms are also racing to fill.

IVManagement

Sam Sledge, 39, has been chief executive since August 2021. He joined in 2011 as a frac technical specialist and rose through operations and strategy.⁵ Adam Muñoz, 43, president and chief operating officer, joined in 2010 to start the Permian pumping operation.⁵ The chairman, Phillip Gobe, 73, was chief executive for 17 months from March 2020 and sat on Pioneer's board until its merger with Exxon.⁵ The finance chief, Caleb Weatherl, 38, joined in July 2025 after running a private Midland oil producer and, before that, the finance side of another.⁵ His predecessor was dismissed without cause in March 2025, and the chief accounting officer resigned effective 30 October 2026, leaving the CFO as interim principal accounting officer.¹⁰²⁵ Two senior finance changes within 19 months, while the company raises $1.5B, is something I watch.

The history needs saying plainly. In 2019 the audit committee reviewed the company's disclosure of fleet purchase agreements, then widened the review. It found about $370,000 of expenses reimbursed to senior executives, including the former chief executive, that were theirs to pay, a related-party transaction by the former chief accounting officer that was not properly disclosed, and weaknesses in controls. It found no material accounting errors and no restatement.⁴ Insurers paid $9.8M in 2021 and $10.4M in 2022 toward the cost of the SEC investigation and shareholder litigation that followed.¹⁴ That was largely a different management and board from today’s, and the company says it has rebuilt controls. It is also why I read the newer filings closely.

How they are paid

Sledge's summary compensation for 2025 was $6.54M.¹⁰ His annual bonus target is 115% of salary and 90% of it rests on set measures: Adjusted EBITDA, free cash flow, safety, and, since 2025, PROPWR committed megawatts, worth 10%.¹⁰ The rest is long-term awards. Half of his March 2025 grant was restricted stock that vests over three years, and half was performance units that pay on ProPetro's three-year total return against 11 oilfield-service peers.¹⁰ I read three incentives in that design. Adjusted EBITDA and committed megawatts pay for earnings and for signing, not for the return on the $1.6B being spent. Relative return against pumping peers pays for share price, not for owner cash. And no measure counts the shares issued. A manager paid this way has a reason to sign contracts and to fund them, with whatever works. I do not say that is what they are doing; I say that is what the pay rewards.

What they own, and what they sold

Sledge owns 649,336 shares, about $5.0M at $7.73 C (calculated).¹⁰ All directors and officers together own 1.55%, about $15M C (calculated).¹⁰ Against pay of $6.5M a year, I would call that a small stake rather than a large one. In the 12 months to 30 October there were no open-market purchases by any insider. Two directors sold: Larry Lawrence 27,000 shares at $10.84 and 35,831 at $11.16, and Michele Vion 6,916 at $10.64; Muñoz sold 17,230 at $10.14 and 70,696 at $11.17.¹¹ The rest of the filings are vesting and shares withheld for tax. These are small sums against holdings, and they are also the prices at which insiders chose to sell, $10 to $11, 40% above today's.

The large seller was Exxon, which sold its entire 16,600,000 shares at $16.66 on 20 May 2026, about $277M C (calculated).¹¹ Exxon is also a customer, and it had held the shares since the 2018 asset sale.² The other five-percent holders in the March proxy were Sourcerock Group (5.58%), Jeal Intec (5.25%), Vanguard (5.23%) and BlackRock (5.05%).¹⁰

What they did with capital

The test of a ten-year cycle is what owners kept. From 2016 to 2025 the company took in $12.5B of revenue, earned $86M of net income in total and generated about $11M of owner cash, after paying for equipment and stock awards C (calculated).⁶ In the good years, 2018 and 2019, cash went into new equipment and the Pioneer deal. In 2023 and 2024 it bought back 13.0M shares for $110.8M, an average of $8.53 a share.¹⁵ In January 2026 it sold 17.25M shares at $10.00 for $163M.² Buying at $8.53 and selling at $10 is not a bad result, and the sale was needed to fund the engines. But the share count is 122.9M now against 103.0M at the end of 2024, and 152.6M if the convertible converts C (calculated).²⁶

VFinancials

ProPetro has audited filings from before its 2017 listing, so there are ten full years of the pumping business and about five quarters of the power business. I show both, and I say plainly that the second is too short to model on its own history.

The pattern is a cycle with no trend. Revenue reached $2,052M in 2019 and fell to $789M in 2020. It came back to $1,630M in 2023 and has slid to $1,269M in 2025 and $1,222M over the past twelve months.⁴ Adjusted EBITDA peaked at $519M in 2019 and was $192M over the twelve months, a margin of 15.7% against 25.3% at the top.⁴ Part of the fall is the lease: from 2023 the company has paid about $63M a year in rent for its electric fleets, and that rent is inside Adjusted EBITDA, where in 2019 the same equipment would have been bought.¹⁵³ Net income over the ten years was $86M in total, against $12.5B of revenue.

What an owner could really take out

Almost nothing, and for ten years. Add up operating cash flow minus capital spending plus asset sales minus stock pay and the 2016 to 2025 total is about $11M C (calculated).⁶ The figure was positive in five years of ten, 2018, 2020, 2021, 2023 and 2024, and negative in the other five. In 2019, the best year for earnings, the company spent $503M on equipment and owner cash was −$48M.⁴ The reason is that a pumping fleet wears out. The company wrote off $58M to $107M of equipment in each year from 2018 to 2022 as losses on disposal, and Adjusted EBITDA adds those back.⁴¹⁵ Replacing worn components is a real cost that the metric hides.

The latest twelve months are worse because the power programme is under construction. Operating cash flow was $196M, capital spending incurred $398M, asset sales $20M and stock pay $21M, so owner cash flow was −$203M C (calculated).⁶²³ Capital spending incurred was $99M higher than what the company paid in cash over nine months, $76M of it because the lender pays the engine maker directly; that is real spending.³

Quality of earnings

Two adjustments matter. Adjusted EBITDA adds back $21M a year of stock pay, which I deduct. It is struck after the $63M of electric-fleet rent, which makes it a conservative number until the fleets are bought.³ Working capital is the other drag: receivables rose by $89M in nine months to $290M, about 74 days of sales.³ Operating cash flow was $115M in nine months against $151M a year earlier.³

Completions earnings were hurt in the second and third quarters by costs the company chose to incur: standing up a twelfth and then a thirteenth fleet, hiring crews, restocking parts. The company says those costs will fall in 2027, and that fourth-quarter results will be seasonally soft.¹⁶⁸ Q4 2025's Adjusted EBITDA was $51M on $290M of revenue, 17.6%, and I assume the fourth quarter of 2026 is near $42M for completions E (estimate).³⁵

The balance sheet

At 30 September the company held $729M of cash against $862M of debt at face value, net debt of $135M C (calculated).³ It had $1,009M of equipment net of depreciation, up from $793M at the end of 2025.³ The cash is not free: remaining equipment commitments were $633M at 30 September, $463M of it Caterpillar's, and the company added a $53M battery order in October.² The company says its raised funds "largely" cover PROPWR's near-term needs.⁸ The next table lists what is owed and what is promised.

The Caterpillar contract deserves a careful read. PROPWR must place orders at least 15 months before delivery, up to minimum amounts set for each year, and if it falls short it pays a fee that the 8-K puts at 75% of the price of the equipment not bought (the exhibit blanks the rate, so the 75% rests on the 8-K alone). Caterpillar has to show the shortfall was not its own failure to deliver or a force majeure event.⁹²⁴ If Caterpillar resells the engines within 12 months, PROPWR gets a credit equal to the fee less 15%, usable on future Cat engine orders, not cash.⁹ PROPWR can end the agreement for convenience, but it still owes the minimums for a notice period that the filing redacts.⁹ The minimums for 2029 to 2031 reset to a percentage of forecast demand on 31 December of 2027, 2028 and 2029, so they follow a forecast the company supplies.⁹ I cannot see the yearly minimums or how much of the $1.1B is already ordered. I know that 570 MW is on order with $463M left to pay, and that the full framework is about 1.5 GW plus an option, so a large part is still to be ordered or paid for by deadlines I cannot see. That is the bear case's main input.

VIValuation

First, how this goes wrong

Before any number, I ask what would most likely hurt an owner, and work back. I see three ways, and they can arrive together.

Pumping falls while the power money is spent. The company has committed about $1.15B of power spending between now and 2028 whatever the oil price does. Completions revenue fell 61% between 2019 and 2020 and 22% from 2023 to 2025, and the stock reached $1.51 in March 2020.⁴¹ If a downturn arrives in 2028, the same year spending peaks, there is no cushion from the pumping side. The power contracts do not arrive. Only 510 MW of 1,100 MW on order is signed, 60 MW of it with a data centre, and 70 MW of earlier contracts have already lapsed.⁷¹⁸ Unsigned engines are paid for, or carry a 75% fee, and resale is not guaranteed.²⁴ The money costs too much. The company must find roughly $0.4B more in my base and $0.6B in my bull, and the $690M convertible falls due in November 2031 unless it converts.²² If the shares are cheap when it needs equity, existing owners pay.

What an owner can really take out

Reported earnings and owner cash are far apart here. Over the twelve months to September, net income was $2M and Adjusted EBITDA $192M, but after paying for equipment and counting stock pay as a cost, owners were $203M worse off C (calculated).⁶²³ Looking forward, my base has owner cash of $-498M in 2027 and $-315M in 2028 as the build is paid for, turning to +$211M in 2029 and about $271M a year from 2032 on 586M of EBITDA, so about 46 cents of each dollar of EBITDA reaches owners E (estimate). The rest goes to stock pay, replacing worn equipment, upkeep on the engines and tax.

What I cannot judge

Four inputs are not knowable from here, and I show a range and not a figure. First, what a megawatt really earns: one quarter of segment results exists, at about $170k a megawatt a year on my guess of the fleet running, against a target of $230k to $375k. Second, whether the data-centre talks produce signed contracts, since 60 MW is the only one in a year. Third, what engines fetch if they are not deployed, which depends on a power shortage that may or may not last. Fourth, how fracturing prices reset when most horsepower renews in the first half of 2027. I would not trust any one-number answer, including mine. The table of cases below is the honest form of the answer.

The discount rate

The ten-year Treasury yield was 4.94% on the price date and Damodaran's implied equity risk premium 4.20%.²⁶²⁷ ProPetro's beta on month-end returns against the S&P 500 is 0.66 over three years, 0.85 over five and 1.96 since 2017, which gives a cost of equity of 7.7% to 13.2% C (calculated).¹²⁸ A range that wide tells me the measure is unreliable, and I do not let the 3-year figure set a rate near 8% for a levered oilfield business building a new line. The convertible is a 0% coupon, the Caterpillar loans cost 7.73%, and the lease line SOFR plus 6.25%.²²³ I use 11% for the business as a whole, which matches a beta near 1.45, between the five-year and since-listing figures, and show 9% to 13%. Cash flows are in nominal dollars.

The three cases

I model the business as two parts on one balance sheet. Completions revenue and margin are set from the company's own history; power is modelled by cohort, megawatts running times what each earns, less upkeep, with the $1.4M to $1.5M a megawatt capital cost, the 1,100 MW already ordered following the known commitment schedule, and any extra megawatts costing the same. Cash tax uses 24%, 100% expensing of new power equipment, and a loss pool. Each year's owner cash is after stock pay, sustaining capital, growth capital and tax.

Bear

Completions goes back to a trough like 2020 and 2021: revenue falls to $1.1B and the margin before lease savings to 13.5% in 2028 and 2029. Only 650 MW of power is ever running, at $225k a megawatt, and the first contracts re-let at 70% of today's price. About 450 MW of engines already bought are sold in 2029 at 75% of an engine cost I put near $1.0M a megawatt and $60M of Caterpillar shortfall fees are paid. The business is worth about $199M, barely more than the net debt, so a share is worth $0.52.

Base

Completions earns a margin of 17% to 17.5% on $1.3B of revenue, below the 2022 to 2024 average, and the electric fleets are bought out. The 1,100 MW of equipment already delivered or on order is all contracted and running by 2030, at $270k a megawatt, a payback of 5.4 years, and re-lets at 80% of the first price. The programme earns its cost of capital and no more. A share is worth $5.84.

Bull

Completions repeats 2023: a margin of 22.5% on $1.5B. The whole 2.6 GW is signed and running by 2032 at $320k a megawatt, a payback under 4.4 years, with contracts that last and re-let at 95%. Peak net debt is $1.4B in 2030. The convertible stays debt because the value stays below $23.17. A share is worth $19.03.

The funding need is the number I trust least. At 30 September the company had $729M of cash, and my base spends $202M of it in the fourth quarter and $0.8B more in 2027 and 2028. Net debt peaks at $1.16B in 2028, about 3 times that year's EBITDA, and takes new borrowing or equity of about $392M beyond today's debt. The revolver ($185M free), the Caterpillar loans ($20M) and the Stonebriar line ($350M) add up to $555M of room, so the base can in principle be debt-funded.² The bull needs $597M. In the bear I assume the company funds about $260M, but a bear is when it would not be easy to find.

If the $392M had to come from shares, the answer depends on the price. Sold at $7.73 it adds value to the existing holders ($6.39 instead of $5.84) because the shares are above my base; sold at $5 it takes the base to $5.52, and at $3.50 to $4.74 E (estimate). The company sold 17.25M shares at $10.00 in January and bought back 13.0M at an average of $8.53 in 2023 and 2024.²¹¹⁵ I think it will avoid issuing below its own view of value, and I do not think it can always avoid it.

What the price says

Turn it round. At $7.73 the market values the shares at about $965M and the business at $1.10B with net debt C (calculated). My base is $864M after taking out $1.15B of power spending still to come. Something has to fill a gap of $236M.

The simplest reading: the market expects each megawatt to earn about $316k a year, a payback of 4.6 years, which is inside management's four to six year range and in the better half of it. That is not an absurd ask. It is one that a single quarter of data cannot confirm. Notice what it does not need: a 2.6 GW fleet.

How much the answers move

Two surprises. A smaller fleet is worth less, not more, because the equipment is already on order. And a much bigger one is worth about the same, because at a 5.4 year payback each additional megawatt earns about its cost of capital. Value comes from returns above the cost of capital, and the whole question is whether the realised payback lands near four years or near six. At four years the base would be about $9.50 a share; at six years, $4.60; at 6.7 years, $3.60.

The margin of safety

I set it at $3. That is 60% of my base value of $5.84, which is $3.50, rounded down to the whole dollar, so it sits at 51% of base. The usual rule, bear value divided by 0.75, is useless here because the bear is about $0.52. I chose a level where the base could be about half wrong, which on this page is a modest allowance given the table above. The shares last closed at $3.50 or less in 2020.¹ The zone is below $3.

The price against the range

The price is 32% above my base and 59% below my bull. At $7.73 my base cash flows return 11.5% a year to the end of 2036, the bull 23.6% and the bear −0.5% E (estimate). Weighting the cases 25%, 50% and 25% gives $7.81, about the price. The skew explains it: the bull is $19 and the bear is about $0.52, and the price is a bet that the megawatts come. The market may be right and I may be wrong.

The 11.5% base return needs one caution. It is the return to the shareholder after the debt is paid, and the debt here is cheap: the convertible carries no coupon until 2031. A shareholder of a business that will carry about $1.2B of net debt on $0.4B of EBITDA would want more than the 11% I use for the business as a whole, so I do not read 11.5% as clearing the bar.

Against the alternative

Every dollar here could go elsewhere. A ten-year Treasury pays 4.94%.²⁶ My base return here is 11.5%. The nearest comparison on this site is Constellation Energy, which owns long-lived power plants that data centres also want, whose price sits 20% above my base value and whose base return is 7.7% a year; and GE Vernova, which builds the turbines, at 5.8%. ProPetro's 11.5% is higher on paper, and that is expected: it is a small company with a single hyperscaler contract, most of its capital still to be spent and a share price that has halved in six months. A higher base return from a riskier business is not a better deal.

Quality and price, separately

How good is the business? The pumping side is ordinary: it makes money at the top of a cycle and loses it at the bottom. The power side is promising and unproven, with a good counterparty in Targa, one running site and a payback that sounds high but earns about the cost of capital once upkeep, tax and a finite life are counted. How good is the price? It is above my base by 32% and on my numbers that leaves no cushion. A fair-to-ordinary business at a price above my estimate of its value leaves the buyer relying on the bull case, and I would put this one in that group, without calling my base certain.

How the odds multiply

The case for the price needs several things to go right at once. My guesses are in the table. They are not independent, since a company that signs data-centre contracts also finds capital more easily, so the true chance is higher than the product. Even so, four 55% to 60% steps leave about 12% for all of them.

Where this could be in 5 and 10 years

Here is my model's value at two later dates. Each is the value at that date of the owner cash that comes after it, less the net debt I roll forward, per today's 124.9M shares and in that year's dollars E (estimate). Owner cash paid out counts whatever net cash exceeds $200M, which I assume is distributed; before that the cash repays debt, including the convertible in 2031.

For the base to be right at five years, three things must happen: 1,100 MW is signed and running by 2030, completions keeps a 17.5% margin before lease savings, and net debt peaks near $1.2B in 2028 and falls below $0.5B by 2031. At ten years it also needs the power contracts to re-let at 80% of today's price and the engines to run through 2036. The base value is higher than today's mostly because the $1.15B of spending is behind it and the cash flows are in front. The most likely way the base is wrong is that each megawatt earns $215k rather than $270k, which takes the value today to $3.60. The bear differs from the base in the megawatts signed and in the downturn; the bull in the payback and in the 2.6 GW.

These are estimates of value, not forecasts of the share price. The market price can sit above or below value for years.

VIIThe case for the price

What most people miss

Three things are specific and not much discussed. First, the data-centre story rests on one 60 MW contract. In October 2025 the company announced 60 MW for a hyperscaler in the Midwest, and that site has run since June.⁷¹⁶ As far as the releases show, the other 450 MW under contract are oil and gas, midstream and industrial, and when Targa signed in September the company said the 510 MW total "also reflects" earlier oil and gas capacity "no longer committed", which is why 350 plus 230 does not make 510.¹⁸ Second, rent on the electric fleets is $63M a year and sits inside Adjusted EBITDA. Buying one fleet out costs $15M to $20M and, if the cost is the same for the rest, five buyouts are worth about $63M of EBITDA for about $90M.³² Third, fracturing is sold out. With 13 active fleets, a mid-70s Permian fleet count and most horsepower renewing in the first half of 2027, the company has a repricing coming at a time when it says supply is tight.⁸ That helps the old business more than the new one gets credit for.

The bull case

In plain words: gas is cheap in the Permian, power is scarce for data centres, ProPetro has engines coming from the one supplier that can deliver them and the oilfield know-how to run them, and its old business is at the bottom of a cycle with a tight market ahead. Sam Sledge, the chief executive, said on 28 October that "Demand for our completion services remains strong and continues to build".³ The company's October presentation illustrates 0.5 GW running at the end of 2027, 1.2 GW in 2029 and 2.6 GW in 2032, and says payback is four to six years.²⁰⁸ If that is met at the low end, my model gives about $9.50 a share.

On the holder side, one fact with its date. Situational Awareness LP, the fund run by Leopold Aschenbrenner, held 1,960,382 shares of ProPetro, valued at $28,111,878, at 30 June 2026, according to its 13F-HR filed on 14 August.²⁹ That is about 1.6% of the shares outstanding now C (calculated). A 13F shows long positions in US-listed shares only, not shorts or holdings abroad, and says nothing about what a fund does next, so I draw nothing from it. I found no named sell-side view or published investor model for ProPetro that I could open, so I cite none.

Against reality

Today the company earns $192M of Adjusted EBITDA, 262 MW are expected running by the end of the year, and the power segment made $7.6M in the third quarter.³ If it signs another 600 MW by the end of 2028, completions holds a 17% margin and each megawatt earns what the company targets at its midpoint, my base is possible: $5.84 a share. The price needs more. It needs about $316k a megawatt a year, or 2.5 points more completions margin every year C (calculated). That is the distance between today's numbers and the case.

What a 10x would take. Ten times $7.73 is $77.30. By then the convertible would have converted, so there are about 155M shares and a market value of $12.0B. At 8 times Adjusted EBITDA, a generous multiple against the 5.7 times the business trades on now, with no net debt, that needs about $1.5B a year, 21% more than my bull case in 2033, which already has 2.6 GW running at a payback under 4.4 years C (calculated). A megawatt would have to earn about $450k a year, a payback of 3.2 years, outside management's range. Seven years from now that is 39% a year. The shares did rise from $1.51 in March 2020 to $18.20 in April 2026, twelve times in six years, but that was from a crisis low.¹ The odds multiply to under 1%: I give 30% to the whole framework being signed and running, 15% to a megawatt earning $450k, 35% to the market paying 8 times, and 50% to completions holding a 22% margin. For a company of this size a tenfold needs both a bull case and a rich multiple, and I would not plan on it.

Where I come down

Two things moved me: the company has real operating improvement in front of it in pumping, where repricing and the fleet buyouts are on a timetable, and PROPWR has delivered a hyperscaler site that works. They did not move my numbers to the price, because a five-year-plus payback earns about the cost of capital once upkeep, tax and a finite life are counted, and because the data-centre market is one contract so far. A second data-centre contract of a few hundred megawatts, a power segment earning $30M a quarter, and a payback shown to be nearer four years would raise my values the most. A fresh share sale below $5 would lower them.

VIIIRisks, and what would change my view

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

• A few customers. The top five customers were 68.2% of 2025 revenue and Exxon 24.9%.⁵ The Exxon sub-agreement for two electric fleets is expected to end in late 2026 without renewal, and the company has to redeploy the fleets.²
• Oil price and rig count. WTI averaged about $97 a barrel in September 2026 against about $58 in December 2025, as the Middle East war disrupted supply, and the Permian rig count rose to 270.² Pumping demand follows both. A reversal would hit the old business while the new one is being built.
• Engine supply terms. A 75% fee on missed minimums, with only a 15%-haircut credit if Caterpillar resells, is a real liability, and many of the terms are redacted.⁹²⁴
• Equipment and execution. PROPWR has run for about a year. Engines fail, sites slip and permits take time. In the second quarter an out-of-basin fleet job suffered "significant unexpected downtime".¹⁶
• Contract quality. Oil and gas terms are short, about five years in the January update, and a customer's fall-away of 70 MW already happened once.¹⁷¹⁸ I cannot see termination rights.
• Competition and the grid. The 10-K says distributed power competes with grid access and expects significant competition as mobile power develops.⁵
• Controls and people. The chief accounting officer leaves on 30 October; the finance chief is interim in that role. The company had material weaknesses in 2019.²⁵⁴
• Dilution and refinancing. The convertible is $690M due November 2031; the January share sale was at $10.00.²²²¹

Geopolitical risk, and what it means for the value

ProPetro pumps in the Permian and its known power sites are in the United States, with no foreign operations or capital controls of its own. The exposure runs through oil, equipment costs, Permian rules and Caterpillar. For each I changed one thing in my $5.84 base.

Oil. WTI averaged about $97 a barrel in September against about $58 in December, after the war between Israel, Iran and the United States, and the Permian rig count rose to 270.² The combatants, OPEC+ and sanctions on Russia decide that, not ProPetro. If oil falls back toward $60 and completions drifts to my bear levels, the base is $3.19. I give that about one in three, since $58 was ten months ago. If oil stays high and fleets reprice, it is $10.99.

Tariffs. The 10-K records a 25% US tariff on steel imports from March 2025 and an announced 10% tariff on imports from almost every country, and says power equipment leans on foreign supply chains.⁵ Caterpillar's price is fixed for shipments to 31 December 2027, then follows a producer price index, and can be adjusted for tariffs.⁹²⁴ So 2028 engines and site equipment are exposed. If equipment costs 10% more and customers do not repay it, the base is $4.23. I give that about one in four: the 10-K says tariff announcements were followed by exemptions and pauses.

Permian rules. The 10-K calls air permits an advantage, warns that rule changes could raise costs, and says regulators in Texas are acting on produced-water disposal and earthquakes, which costs customers.⁵ A one-year delay on every new megawatt, with engines still paid for, gives $5.28. I give it about one in four.

Gas and power. Customers buy the gas, so ProPetro has no direct gas price risk. Scarce grid power is why anyone rents an engine farm, and the company says gas-burning fleets are sold out on the diesel and gas gap.⁸ At $300k a megawatt a year the base is $7.06, near the price. I give that about one in three.

Caterpillar. Supply is the edge and the exposure. A one-year slip in deliveries gives $5.56, since cash goes out later too. If only 800 MW is signed, 300 MW of engines are resold at 75% and a $60M fee is paid, it is $4.74, about one in four. The exhibit blanks the fee rate; 75% is from the 8-K.⁹²⁴

In the price: $7.73 needs $316k a megawatt, so it leans to the favourable rows. I do not see a fall in oil, a tariff step or a permit delay in it. The cases overlap; I do not add them.

ProPetro short reports: Culper Research, 2019

Who and what. Culper Research is a short seller. Its 2019 reports say it is short ProPetro, so it may profit if the price falls.³⁰ It published three: “Friends & Family First at this Permian Cesspool” (31 October 2019), “There’s No Fracking in Cabo” (4 November) and “Q3 Disclosures Cast Further Doubt on Internal Review” (14 November).³⁰³¹³²

What it alleges. The first report says the company’s 9 October claim that its review was substantially complete was not credible (p. 2). It cites management’s role in the review, the audit committee’s independence, and at least nine undisclosed outside entities tied to executives, staff and board members (pp. 2 to 5). It also calls the company a restructuring story and says the shares are “ultimately worthless” (pp. 2 and 17).³⁰ The second lists 59 private-jet trips it could not easily explain as business (p. 3).³¹ The third says a $3.6M related-party item disclosed on 13 November shows the review fell short (p. 2).³²

The company’s response. I found no reply from ProPetro to Culper by name, and no filing I read mentions it. Culper says Bloomberg got no comment (second report, p. 2).³¹ On 9 October the company said fact-finding was substantially complete and nothing found so far required a restatement.³³ On 13 November it disclosed one related-party transaction it had not reported, about $3.6M paid, linked to its former chief accounting officer, and at least two material weaknesses.³⁴

What is established. The company’s review, reported in its 2019 10-K of 22 June 2020, found about $370,000 of managers’ expenses wrongly charged to the company and later repaid, one undisclosed related-party transaction, and control failures, including that it “did not appropriately identify and monitor conflicts of interest.” It found no material accounting error and no restatement. On 13 November 2019 the company also told investors not to rely on its 2018 report that controls were effective.⁴³⁴ On 22 November 2021 the SEC settled with the company and its former chief executive, who both agreed to cease-and-desist orders without admitting or denying. The order finds proxy statements filed in 2018 and 2019 understated his perks for 2017 and 2018 by $142,570 and $285,615 and left out two pledges of his shares, and that the company violated reporting, books-and-records, internal-controls and proxy rules. It imposes no penalty on the company, citing its cooperation; he agreed to pay $195,046.³⁶³⁷ Neither the release nor the order mentions Culper.

What is not established. No court or regulator has ruled on whether the review was independent. The order says the audit committee’s outside counsel uncovered the improper reimbursements and the pledges (paragraph 30). It finds the company paid about $252,896 for the former chief executive’s personal flights in 2017 and 2018; it does not test Culper’s 59 trips.³⁷ The class action’s third amended complaint (30 July 2020) lists Culper’s first report among the events behind alleged price falls; the settlement carried no finding of fault.⁴³⁴²

Where it touches my model. A related-party leak would sit in completions margin, where one point is worth about $0.76 a share E (estimate) (reverse DCF) and $13M a year on my $1.34B of revenue C (calculated); the SEC’s largest annual perks figure is about 2% of that.³⁷ The other input is the discount rate: my 11% has no governance charge, and one more point takes my base from $5.84 to $4.75 E (estimate). The $24.2M of 2019 review fees and the insured settlement are past and not in my cash flows.⁴⁴⁰ Revenue, share count and capital spending are untouched. I cannot judge from outside whether the other entities mattered.

So what. The SEC and the company’s own review found undisclosed perks and pledges and weak controls in 2017 to 2019, small beside 2018 Adjusted EBITDA of $388M⁴, so my values do not move. A problem later settled is not a finding about today’s company, and a fixed problem does not show there are no others. For Culper’s view to hold today, related-party dealings of size would have to be hidden and the controls missing them; no court or regulator has tested the other entities, and I cannot from outside. For the company’s account to hold, disclosure and controls must now work; clean audit opinions for 2020 to 2022, 2024 and 2025 point that way, and the 2023 manual-journal-entry weakness shows lapses happen. Culper said in 2019 that it was short and today’s executives are paid partly in stock, so each side has a motive to frame it; the next check is the 2026 10-K, due by mid-March 2027 (the 2025 one came on 19 February).⁵