Baker Hughes is a high-quality LNG/gas-turbine franchise (IET) yoked to a cyclical, capital-intensive oilfield services business (OFSE). At $69 the market is paying for the LNG bull case and giving you the OFS optionality almost free, but the 10-year ROIC of -34.8% says the consolidated business has not yet proven it compounds.

Plain English

Baker Hughes does two things. It makes giant gas turbines that liquefy natural gas and power factories and data centers — that part is good, with only two real competitors. And it sells drill bits and pumps to oil companies — that part is okay but commoditized. The company spent ten years cleaning up a bad merger with GE, so its long-term return on capital looks terrible. The recent numbers are better. It just agreed to buy Chart Industries for $13.6 billion, which is expensive but strategically sensible. The stock is fairly priced for the good case and cheap for the great case.

Thesis

Thesis

Baker Hughes (BKR) is two businesses sharing a balance sheet. Industrial & Energy Technology (IET) designs and services aero-derivative and heavy-duty gas turbines that liquefy natural gas and increasingly power data centers — a genuine oligopoly with Siemens Energy and Mitsubishi, with multi-decade installed-base aftermarket annuities. Oilfield Services & Equipment (OFSE) is a commoditized #3 player behind SLB and Halliburton, selling drill bits, pressure pumping, completions, and subsea systems on a well-by-well basis to upstream E&Ps.

The scorecard composite is 72/100 — respectable, but the components are noisy. Profitability scores only 13 and the 10-year average ROIC is -34.8%, dragged by the GE Oil & Gas spin-merger goodwill writedowns and pandemic-era impairments. The forward-looking story is better: 5-year ROIIC is 205% (very high incremental returns as IET scales), TTM owner earnings of $2.77B, and net debt is negative at -12.3x EBITDA (cash-rich, even after issuing notes for the $13.6B Chart Industries acquisition closing Q2 2026).

The scorecard's intrinsic value range is $66.17 / $119.63 / $155.29, against a $69.12 share price — a px/IV ratio of 0.58. That is genuine margin of safety if you accept the base case, but the scorer flags the base CAGR was clamped from 87% to 14% and maintenance capex has >50% spread, so the IV high is fragile.

The reverse-DCF implies just 6.87% growth is priced in. That is plausibly beatable given $3B in data-center turbine orders booked 2025-2027 and Chart's cryogenic equipment locking in molecule-handling share. The price/IV math says: if base case is right, you make ~75% to fair value; if low case is right, you are flat. Acceptable asymmetry — but only if you trust the IET franchise to carry a structurally inferior OFS business through the next downcycle.

Moat

Moat Assessment

Pricing power — Mixed. IET's gas-turbine franchise enjoys real pricing power: aero-derivative LFA-class turbines for LNG trains are a three-way oligopoly (Baker Hughes, Siemens Energy, Mitsubishi Heavy Industry). Once specified into a $5-15B greenfield LNG plant, the OEM is locked in for 25+ years of parts and overhauls — the classic razor/blades pattern. The 2025 10-K discloses long-term equipment contract revenue and long-term product service agreement revenue as separately tracked deferred revenue lines, evidence the aftermarket annuity is meaningful. OFSE has essentially no pricing power: contracts are awarded well-by-well, and the 10-K explicitly notes 'principal competitive differentiators... are technology, quality, efficiency, reliability, and availability' alongside 'competitive pricing.' That is a polite way of saying it competes on price every cycle.

Switching costs — Real in IET, weak in OFSE. Once a Frame 5/7/9 turbine or LM2500/LM6000 aero is installed at an LNG plant, swapping to a competitor mid-life is essentially impossible — you'd have to pull the machine, re-permit, re-pipe, re-train operators. Bently Nevada vibration sensors are similarly embedded in customer plant DCS systems. Cordant AI software, if it gets traction, becomes a further switching-cost layer. In OFSE, switching costs are near zero — operators run multi-vendor strategies and rotate suppliers each AFE.

Network effects — None of consequence. The data assets across Cordant and Waygate Technologies could theoretically benefit from scale (more inspections → better defect models), but this is a stretch versus a true two-sided network.

Intangibles — Moderate. $600M R&D and 1,400 patents granted in 2025 are real but modest for a $30B+ revenue base. Brand strength matters at the margin (Bently Nevada in vibration monitoring; Druck in pressure sensors; Waygate in NDT) but customers are sophisticated industrial buyers who run RFPs.

Cost advantages — None proven. This is the big tell. The 10-year average ROIC of -34.8% is the empirical answer. Even adjusting for one-time impairments, BKR has not earned its cost of capital across a full cycle. Unlike SLB (the cost leader in OFS) or GE Aerospace (the cost leader in aero-derivative turbines for aviation), BKR sits in the middle on both businesses.

$10B / 5-year competitor stress test. If a Chinese state-backed turbine OEM (Harbin, Dongfang, or Shanghai Electric) committed $10B over 5 years to attack the LNG turbine market — at subsidized cost-of-capital and with Belt-and-Road financing — IET's moat would erode in non-Western markets within 5-7 years. Already CNPC and Sinopec have specified domestic turbines for sub-critical applications. The Western installed base is safer due to FCPA, ITAR, and ESG scrutiny, but the growth markets (Africa, Southeast Asia) are vulnerable.

Erosion risks. (1) Long-term LNG demand erosion if heat pumps, grid storage, and SMR nuclear undercut gas in OECD markets — Buffett warns repeatedly that 'past universe' models mislead in regime changes [1][2]. (2) Climate Technology Solutions (CCUS, hydrogen) are speculative — Buffett's 1984 letter warns that managers fund unprofitable bets to stay busy. (3) The Chart Industries deal at $13.6B is a price-paid risk: integration on a Lubrizol/Precision Castparts timetable [1] could trap returns for years.

Moat verdict: NARROW — wide on the IET installed-base annuity, none on OFSE; the consolidated business deserves NARROW.

Management

Management & Capital Allocation

CEO Lorenzo Simonelli has run BKR since the 2017 GE Oil & Gas / Baker Hughes merger. He inherited a structurally damaged business — the GE-era goodwill produced the cratering 10-year ROIC of -34.8% — and has spent eight years cleaning it up. His record on the five capital allocation choices:

1. Reinvestment. $600M R&D in 2025 is roughly 2% of revenue, which is light for a 'technology company' but consistent with the industrial OEM peer set. The 5-year ROIIC of 205% says incremental dollars are earning real returns — though the scorer notes this 5-year window is post-impairment cleanup, so it flatters management. Verdict: B.

2. M&A. Three deals matter. (a) Chart Industries — $13.6B announced July 2025, expected close Q2 2026. At ~$210/share, ~12-15x forward EBITDA on Chart's cryogenic and process equipment business. This is not cheap, and the 10-Q discloses a special-mandatory-redemption clause on the new senior notes — a tell that management itself sees the regulatory risk as non-trivial. (b) Continental Disc Corporation added rupture discs and actuators to Industrial Products — small, sensible bolt-on. (c) PSI divestiture to Crane Company (closed Jan 2026) — selling Panametrics/Druck/Reuter-Stokes for industrial focus. The portfolio surgery is rational; the Chart price is the bet to watch. Verdict on M&A: C+ — too soon to grade Chart, which dominates the next three years' returns.

3. Debt. Net debt is -12.3x EBITDA (i.e., net cash), but this is pre-Chart. After Chart closes, BKR will absorb ~$13.6B cash + assumed debt, swinging to modestly levered. The 10-Q shows new senior notes issued in Q1 2026 explicitly to fund Chart, with a special mandatory redemption if the deal breaks. Investment grade should be preserved. Verdict: B.

4. Buybacks. A repurchase program exists but is unspecified in size and is being deprioritized to fund Chart. Historical buyback discipline (avg P/IV when buying) is unclear — the scorer notes share count change over 10 years is unavailable. This is a yellow flag: BKR has occasionally bought back stock at multi-year highs. Verdict: C.

5. Dividends. Modest dividend (~2% yield), grown gradually. Reasonable. Verdict: B.

Communication quality. Investor-day cadence is regular, segment disclosure is clean (OFSE / IET split with sub-product-line orders), and the 10-K is unusually candid about LNG demand sensitivity. The company quantifies data-center order book ($1B booked 2025, $3B target 2025-2027) — specific and falsifiable, the right kind of disclosure. Verdict: B+.

Buffett/Munger-style red flags to watch. (1) Capex pulled forward to chase the LNG / data-center wave — easy to mistake cyclical peaks for structural growth, the Buffett 2008 warning about 'history-based models' [3]. (2) Chart purchase price discipline — paying ~12-15x EBITDA for cryogenics at a moment when LNG capex is at cycle peak. (3) Climate Technology Solutions (hydrogen, CCUS) is a 'busy work' segment with negative FCF that requires Munger's skepticism toward managers funding unprofitable optionality. (4) Long-tail product service agreements create deferred-revenue accounting room for aggressive estimation — Buffett's 1984 insurance-reserves caution applies analogously to long-cycle equipment service contracts.

Capital allocator: B-

Simonelli has been a competent steward but is now placing the largest bet of his tenure (Chart) at a moment of cyclical exuberance in the energy-tech segment. Grade is conditional on Chart integration discipline.

Industry

Industry Structure — Porter's Five Forces

1. Threat of new entrants — LOW for IET, MEDIUM for OFSE. Building a heavy-duty gas-turbine business from scratch is a 30-year, $20B+ undertaking with regulatory, certification, and metallurgy barriers (single-crystal blades, hot-section coatings). Effectively only Siemens Energy, Mitsubishi, and BKR/GE-lineage compete in the LNG-class machines. Chinese SOE entrants exist but are confined to domestic markets. OFSE has lower barriers — Weatherford, NOV, TechnipFMC, and dozens of regional players churn in and out each cycle.

2. Bargaining power of suppliers — MEDIUM. Specialty alloys (Inconel, single-crystal nickel superalloys), forgings, and electronics are concentrated supply chains. Precision Castparts (Berkshire-owned) [1] is a major supplier to the turbine industry, giving the supplier side real pricing power. Rare earths and Chinese-sourced inputs are a strategic vulnerability.

3. Bargaining power of buyers — HIGH on OFSE, MEDIUM on IET. OFSE customers are super-majors (Exxon, Chevron, Shell, Aramco, ADNOC), national oil companies (Petrobras, CNPC), and US shale independents — all sophisticated, multi-vendor buyers who extract concessions every downcycle. IET customers (LNG project sponsors, midstream operators, data-center hyperscalers) are also large but face fewer alternative turbine OEMs, giving BKR more leverage. Customer concentration is meaningful: the 10-K cites US and UAE as concentration-risk geographies for receivables.

4. Threat of substitutes — MEDIUM, rising. For OFSE: shale efficiency improvements and electrification of drilling rigs reduce service intensity per barrel. For IET turbines: SMR nuclear, grid-scale battery storage, and hydrogen-fuel-cell power are credible 10-20 year substitutes for gas-fired peakers. LNG demand itself is the substitution risk: heat pumps and renewables-plus-storage can displace gas heating and gas-fired electricity in OECD economies.

5. Competitive rivalry — HIGH on OFSE, MODERATE on IET. OFSE is a price-war battlefield each downcycle — SLB's scale and Halliburton's North American presence pressure BKR's #3 position. IET is more disciplined: Siemens Energy and Mitsubishi compete on technology and execution rather than price, and the order book runs 18-36 months out, allowing rational capacity planning.

Value pool location and trajectory. The value pool is migrating from OFSE into IET. In 2025, IET's order book grew on the back of LNG (Qatar, US Gulf Coast, East Africa) and a new pillar — data-center power generation. The $1B of data-center orders booked in 2025 and the $3B target through 2027 represents a structurally new customer base (hyperscalers) that did not exist five years ago. OFSE, meanwhile, faces secular pressure from US shale efficiency gains (more oil per rig, fewer rigs needed) and electrification.

Industry Verdict: Good (consolidated). IET alone would be Excellent; OFSE alone would be Average-to-Poor. The blended business benefits from cross-cycle diversification but is held back by OFSE's commodity dynamics.

Inversion

Inversion — The Bear Case

I am now a short-seller. My job is to articulate the strongest credible case that BKR is significantly overvalued at $69 and could be worth materially less within five years.

1. The single event that kills this

A US LNG export demand reset. The base bull case rests on US LNG export capacity reaching ~30 Bcf/d by 2030, with BKR providing turbines, compressors, and (post-Chart) cryogenic heat exchangers for the bulk of greenfield trains. The single event that kills this is a combined shock: (a) European gas demand falls 15-20% as heat pumps, building electrification, and renewables-plus-storage scale faster than IEA models project; (b) Chinese LNG imports plateau as domestic shale and pipeline gas from Russia (Power of Siberia 2) substitute; (c) the Trump administration or successor pauses LNG export permits to defend domestic gas prices for industrial users. Any one of these stalls FIDs (final investment decisions) on Wave 3 LNG projects. Two of them happening together in 2027-2029 cuts BKR's IET orderbook by 40-60% within 18 months. The data-center narrative is too small ($3B over 3 years on a $30B revenue base) to fill the hole.

2. Why the moat is narrower than bulls think

Bulls point to the gas-turbine oligopoly. But the consolidated moat is much weaker than the IET-only moat. OFSE represents roughly 60% of revenue and has structurally inferior economics — it competes well-by-well, has near-zero switching costs, and is the #3 player behind SLB and HAL. The 10-year ROIC of -34.8% is the empirical proof that the blended moat does not exist — adjusting generously for one-time impairments still leaves you below cost of capital. Even within IET, the moat is contestable: Siemens Energy is recovering from its turbine-quality crisis and pricing aggressively for share; Mitsubishi has a stronger Asian-utility footprint; and Chinese OEMs are entering the sub-critical industrial gas-turbine segment in Belt-and-Road markets. The Cordant / Bently Nevada digital-software pivot is a 5+ year bet that has not yet shown stickiness — and competitors (Emerson AspenTech, Aveva-Schneider) have larger software franchises.

3. Why management is worse than it appears

Lorenzo Simonelli has spent eight years cleaning up GE's mess; the cleanup is now mistaken for compounding. The most consequential capital-allocation decision of his tenure — the $13.6B Chart Industries acquisition — is being made at the peak of the LNG and data-center capex cycle. This is the textbook capital-cycle blunder Buffett warns about: 'If merely looking up past financial data would tell you what the future holds, the Forbes 400 would consist of librarians' [3]. Worse, BKR is paying ~12-15x EBITDA for Chart, a price that requires LNG capex to remain at peak levels indefinitely to clear the hurdle rate. The special mandatory redemption clause in the new senior notes is the lawyers' admission that even management thinks the deal could break. The Climate Technology Solutions segment is a slow-bleed 'busy work' bet — Buffett's 1984 'walking dead' insurance metaphor [from canon] applies: keep writing the business to look productive, even when the unit economics don't clear. Management compensation is tied to revenue and segment EBITDA growth — both of which are inflated by Chart and by booking long-cycle equipment orders before profitability materializes.

4. What bulls are extrapolating that won't hold

(a) The 5-year ROIIC of 205% is being extrapolated as if it represents structural quality. It does not — it represents post-impairment recovery from a depressed base. Normalizing for the GE-era goodwill writedowns, true through-cycle ROIIC is probably 8-12%, not 200%. (b) Data-center orders. $3B over three years is roughly 3% of revenue — a real but small contributor. Bulls are pricing in $10-20B/year of data-center demand by 2030, which assumes (i) gas turbines remain the only scalable fast-deploy power source, (ii) hyperscaler capex doesn't reset, (iii) SMR nuclear doesn't ship in 2028-2030. All three are contestable. (c) LNG aftermarket annuity. The bull case treats the installed base as a perpetual royalty. But aftermarket revenue is competed by independent service providers (Sulzer, EthosEnergy, third-party shops), and on older Frame-class turbines third-party service has captured 20-30% of parts revenue.

5. Valuation trap — multiple compression / regime change

BKR currently trades at 23.4x TTM P/E, well above the 10-15x range typical for a blended OFS-plus-industrial business. The IV-base of $119.63 assumes 14% revenue CAGR (the scorer clamped it down from a deceptive 87%) and the 12/17/22 P/FCF multiples — neutral defaults because no historical multiple is available. If Chart integration disappoints, OFS rolls into a downcycle in 2027-2028, and the data-center narrative deflates, you get all three of: (a) earnings reset 25-35% lower as bookings slow, (b) multiple compression to 12-14x P/E, (c) write-down of Chart goodwill. That math: $2.77B owner earnings × 0.7 normalization × 13x = $25B equity value, vs. ~$70B today. If I am right, the stock could be worth $35-40 within 3-5 years.

If I am right, the stock could be worth $38 within 4 years.

Lollapalooza Bias Check

Lollapalooza — Biases Active in Me Right Now

Recency bias. I am writing this in May 2026 after two years of LNG-tightness narratives, post-Ukraine European gas anxiety, and the AI/data-center power frenzy. Every headline reinforces 'gas turbines are the answer.' This narrative is recent, persuasive, and quantitatively backed — exactly the conditions under which recency bias overshoots. I should anchor on through-cycle averages, not 2024-2026 peak metrics. The 5-year ROIIC of 205% in particular is recency-bias bait; through a full cycle the real number is likely a fraction of that.

Anchoring. I am anchoring on the scorer's IV-base of $119.63 as if it were a real number. It is the output of a deterministic Python model with neutral 12/17/22 P/FCF multiples and a clamped 14% CAGR assumption. The scorer notes explicitly flag 'Maintenance capex uncertain (>50% spread); widen IV range' and 'no historical P/FCF available.' I should treat the IV range as a fairly weak prior and lean toward the IV-low of $66.17 — which is essentially the current price. That changes the trade entirely.

Confirmation bias. I have written the 10-K's bullish case (LNG, data centers, IET aftermarket) more vividly than the bearish case (OFS commodity dynamics, Chart price-paid risk, hydrogen/CCUS losses). I had to deliberately force the inversion to be tougher than instinct allowed.

Authority bias. I am giving credit to Lorenzo Simonelli's tenure because BKR is widely considered 'cleaned up.' But the 10-year ROIC is still -34.8%, full stop. Authority of consensus narrative doesn't change the empirical record.

Commitment / consistency. Once I called the moat 'NARROW,' I felt pressure to keep the recommendation positive to remain consistent with the px/IV ratio of 0.58. The honest read is that NARROW moat plus uncertain IV plus peak-cycle M&A risk doesn't combine to a Buy.

Incentive bias (in management, not me). Simonelli's compensation is tied to revenue growth and bookings — exactly the metrics most flattered by an oversized acquisition at cycle peak. This biases the Chart deal toward closing even if price discipline argues against it.

Deprival super-reaction (in the market). LNG turbines are perceived as 'scarce capacity.' This drives buyer urgency and order-book inflation. When perceived scarcity reverses (e.g., SMR nuclear ships, or LNG FIDs pause), the deprival reaction inverts and orders cancel disproportionately — the same lollapalooza effect that drove the spike, but in reverse.

Net effect on my analysis. The biases push me toward Buy. Correcting for them pulls me toward Hold with a tight target buy price below current — i.e., demand a real discount before adding rather than chasing the LNG narrative.

10-Year Outlook

10-Year Outlook Test

Same fundamental business model in 2036? Partly. Gas turbines for LNG and power generation will still exist, and the installed-base aftermarket will still be a real annuity — that part of the business model is durable. But 'oilfield services' as practiced today (drill bits, pressure pumping, wireline) is more uncertain in a decarbonizing 2036 world. Production-side OFS work (subsea, artificial lift, chemicals) will persist; exploration and drilling-services intensity may shrink as oil demand peaks somewhere in 2028-2032 (per IEA, BP, Shell scenario analyses).

Customer base larger? IET customer base is broader (data-center hyperscalers, industrial power-gen, hydrogen) but possibly less deep per customer. OFSE customer base is narrower — fewer, larger NOCs and majors as US shale independents consolidate.

Profit per customer higher? In IET, yes — long-term service agreements with hyperscalers and LNG operators carry better margins than transactional turbine sales. In OFSE, no — pricing pressure from SLB's scale and electrification of drilling reduce service intensity per well.

Moat wider? If Cordant and Bently Nevada digital franchises take hold, the IET moat widens via switching costs in plant operating systems. If they don't, the moat is roughly stable. OFSE moat will not widen — that segment is structurally a cost-plus commodity service.

Single biggest threat. A coordinated demand shock: OECD gas demand reset (heat pumps + renewables-plus-storage) combined with SMR nuclear shipping at scale 2030-2034. This collapses both BKR's LNG turbine orderbook and the data-center gas-turbine narrative simultaneously.

Confidence assessment. The IET franchise is genuinely understandable and likely durable; the OFSE business is cyclical but predictable; the Chart integration is a wildcard; and the LNG demand trajectory has a wide cone of uncertainty (nuclear, electrification, geopolitics). The blend is comprehensible but the two-business structure means I cannot underwrite confidently which path BKR takes. I can imagine BKR worth $35 in a downside case and $200 in a deep LNG / data-center bull case — that 6x range is a yellow-flag-level cone.

CONFIDENCE: medium

Position Guidance

Position Guidance

• Recommendation: Hold
• Conviction: medium
• Target buy price: $55 (a real margin of safety below the IV-low of $66.17, accounting for OFS cyclicality and Chart integration risk)
• Target trim price: $130 (above the IV-base of $119.63 but below IV-high of $155.29, locking in gains before the bull case is fully priced)
• Position sizing: 1-2% of portfolio if entered at $55 or below. Avoid initiating at current $69. The two-business structure (high-quality IET + commodity OFSE) and pending Chart deal make this a sizing-disciplined position, not a high-conviction core holding.
• Catalysts to monitor: Chart deal close (Q2 2026), Wave-3 LNG FIDs (2026-2027), data-center order book growth (target $3B by 2027), OFS rig count direction, SMR nuclear deployment news.
• Re-rating triggers down: OFS downcycle, Chart integration write-down, LNG FID pause, hyperscaler capex reset.
• Re-rating triggers up: Multi-year LNG order visibility, data-center orderbook surprise, IET margin expansion above 20% segment EBITDA.