Energy / Utilities · Deep dive

PG&E Corporation

The 1905-vintage California IOU whose stock lost roughly 20% in a single session on 31 August 2026 when the legislature let SB 492 die without the wildfire-subrogation shield the utility needed — repricing PG&E's cost of capital to include the open-ended tail risk that AB 1054's $21B fund was never designed to cover.

at risk

The 31 August 2026 California wildfire-liability-cap failure just re-priced PG&E's cost of capital to include open-ended tail risk that AB 1054 cannot cover, which is the structural risk no undergrounding spend can permanently hedge.

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

PG&E Corporation is the Oakland-headquartered holding company for Pacific Gas and Electric Company, the largest US investor-owned utility by customer count — ~5.5M electric and ~4.5M gas customers across 70,000 square miles of northern and central California. The equity story going into September 2026 was the AI data-center load-growth thesis: a 12.7 GW interconnection pipeline at 30 June 2026 (up from 5.1 GW at 31 March 2026), a $73B five-year capital plan through 2030, reaffirmed 2026 non-GAAP core-EPS guidance of $1.64-$1.66, and 9%+ EPS growth projected 2027-2030 (Q2 2026 release, 24 July 2026). The equity story going out of September 2026 is something very different: on 31 August 2026 the California legislature let SB 492 die on the final day of session without a vote, stripping the Newsom-negotiated insurance-subrogation shield the utility had priced in, and PCG fell ~20% in one session — the worst day since 18 March 2020. Mizuho, BMO and Wells Fargo downgraded the next morning. On 2 September 2026, the board announced a Strategic Review Committee and a ~$2B deferral of 2027 capex. A company that spent the entire Poppe era rebuilding an investment-grade balance sheet suddenly re-entered the market’s mind as a policy-dependent bond proxy with an open-ended tail.

Founding story

PG&E is genuinely 1905 — one of the oldest continuously operating IOUs in the US. Founders John Martin and Eugene de Sabla, Jr. were Yuba-River gold-mine operators who electrified their claims with hydro in the 1890s. In 1903 they formed California Gas & Electric Corporation as a roll-up vehicle; in 1905 they bought San Francisco Gas & Electric and on 10 October 1905 merged the two to form Pacific Gas and Electric Company, capitalised at ~$45M (Encyclopedia.com; FundingUniverse). The founding operating logic was the physical arbitrage between Sierra hydro and San Francisco steam-peaker capacity — still the topology of the modern system. The 1906 San Francisco earthquake nearly destroyed the six-month-old company; PG&E rebuilt and spent the next century compounding as the default utility for northern California. In 1997 the parent-subsidiary structure was formalised with PG&E Corporation as HoldCo and Pacific Gas and Electric as OpCo — a structure whose purpose was to house non-utility ventures at HoldCo while ring-fencing the regulated OpCo, and which survived both Chapter 11 filings because both were OpCo-only.

The two bankruptcies frame everything. The 2001 filing was a market-design failure: California’s mid-1990s deregulation combined with drought and Enron manipulation to send spot power to $330/MWh while retail rates were frozen, and Pacific Gas and Electric filed on 6 April 2001 with $9B of debt (Power Engineering). The 2019 filing was an operating-safety failure: PG&E equipment sparked the 2015 Butte, 2017 Wine Country/Tubbs, and 2018 Camp Fire disasters that killed 100+ people and destroyed 25,000+ structures, and the ~$30B estimated liability drove the 29 January 2019 second Chapter 11. Both endings shape today’s governance — CPUC enhanced oversight, AB 1054 fund participation, and a board (chaired by Kerry Cooper since 31 October 2024) explicitly organised around safety-and-affordability optics.

How it works

PG&E is a standard cost-of-service IOU with California modifications. OpCo files a general rate case with the CPUC on ~four-year cadence; a separate FERC transmission rate case (TO21, approved 5 August 2025) sets the transmission ROE. Each case argues for a forecast rate base — invested capital net of depreciation and deferred tax — plus authorised return on that base plus recoverable O&M and depreciation. CPUC-authorised ROE is around 10.28% on a 52% common-equity structure (10-K FY 2025); FERC TO21 base ROE is 10.38% on a fixed 50% common / 49.7% long-term-debt / 0.3% preferred structure.

Rate base is the load-bearing number. Electric and gas distribution plus electric transmission plus gas transmission and storage sits at ~$60B (2025 disclosures) and compounds at ~10% annually under the $73B five-year capex plan. Every dollar of rate-base growth times authorised ROE is incremental earnings — that is the entire equity thesis. The plan requires proportional equity issuance to preserve the 52% equity ratio, which is why the story is so sensitive to cost-of-capital shocks like SB 492.

Layered on top are three California-specific artefacts. Each IOU files a three-year Wildfire Mitigation Plan with the Office of Energy Infrastructure Safety, and the safety certificate contingent on WMP approval gates access to the AB 1054 wildfire fund. Public Safety Power Shutoffs (PSPS) pre-emptively de-energise circuits during red-flag wind events. And AB 1054 itself is a $21B state fund funded evenly by shareholders (via a non-bypassable charge) and ratepayers, capped at $10.5B on the shareholder side, that PG&E/SCE/SDG&E can draw on for wildfire liability above $1B per event subject to a prudent-management standard. The SB 492 shock hits one specific spot: the fund is not open-ended and once drained is not automatically replenished. Newsom’s draft would have added an insurance-subrogation shield preventing insurers from suing utilities to recover payouts to fire victims — the exact litigation vector that produced the largest single-event liabilities in the 2015-2018 cycle. Without the shield, the residual tail sits on PG&E’s balance sheet.

Product and business overview

Electric distribution and transmission (~70% of revenue). ~5.5M accounts; ~107,000 miles of distribution and ~18,000 miles of transmission from Eureka to Bakersfield. Generation is a mix of Diablo Canyon nuclear (extended to at least 2030 under SB 846), hydro, gas and contracted renewables. Natural gas (~30% of revenue). ~4.5M gas accounts; backbone gas transmission plus ~50,000 miles of distribution main. The 2010 San Bruno explosion (8 killed) is the operating event that defines this segment’s governance posture — federal 6-count felony conviction in 2016 plus $1.6B CPUC fine and ~$500M civil settlements. Undergrounding programme. The Poppe-era differentiator: 10,000-mile pledge from July 2021; 1,000 miles completed October 2025 (TD World); ~1,600 miles targeted by end-2026 and a claimed ~18% system-wide wildfire-risk reduction at completion. Data-center interconnection. The growth pocket: 12.7 GW pipeline at 30 June 2026; ~1.5 GW in final engineering targeted to energise 2026-2030, concentrated in San José and the Bay Area.

Business model and pricing

Regulated retail revenue is authorised revenue requirement divided across the customer base via bundled residential, TOU, EV, medical-baseline and C&I tariffs. Average bundled residential rates were ~39.25¢/kWh at March 2026 (Solar.com), against a California state average of 34.74¢/kWh (Electric Choice, September 2026) — roughly 89% above the US average of 18.34¢/kWh. Starting March 2026 PG&E rolled out a $24/month Base Services Charge with an offsetting $0.05-$0.07/kWh reduction to the volumetric rate — an explicit shift of fixed-cost recovery away from the volumetric arbitrage surface that rooftop-solar plus battery competition uses. Non-bypassable charges — public purpose, wildfire fund, PCIA (Power Charge Indifference Adjustment for CCA customers), DWR bond charge — sit on top of the base rate and are the mechanism by which departing CCA load still contributes to PG&E’s stranded-cost recovery.

Traction over time

Two things dominate. First, the compounding-machine narrative was intact through Q2 2026 — rate base at ~10%, EPS on track for 9%+, data-center pipeline more than doubling in a quarter. Second, that narrative was hostage to the wildfire cost-of-capital assumption, and 31 August 2026 changed the discount rate. Mizuho’s cut from $21 to $16 is a 24% haircut to implied fair value; BMO’s cut from $28 to $21 is 25%. The 30-day drawdown of ~24% is roughly consistent with ~150-200 bps of implied cost-of-equity added to the same day-forward numbers.

Market analysis

US regulated electric utilities are a ~$500B annual revenue sector that ran at ~0.5% organic demand growth pre-2023 and has been revised to ~2-3% for 2025-2030, driven principally by hyperscale data-center load, secondarily by EV charging and industrial electrification. California is the sharpest expression — PG&E’s 12.7 GW data-center pipeline is roughly 50-60% of current summer peak. Three structural forces run in different directions: decarbonisation and electrification grow rate base and utility earnings (tailwind); wildfire, storm and climate insurance costs grow both opex and tail liability (headwind); and distributed generation, storage, VPPs and CCAs erode captive load and monopoly rents (flank). California leads on all three, which is why PCG and EIX trade at compressed multiples relative to Duke, NextEra or Southern despite comparable growth prospects.

Competitive intel

PG&E’s monopoly on wires and pipes is intact by regulation, but almost every layer above and below is contested. Edison International and Sempra are the peer IOUs — direct beta comparables under the same AB 1054 framework; EIX moved almost lockstep with PCG on 31 August 2026 (-23%). Berkshire Hathaway / PacifiCorp is the extra-territorial precedent for what happens when a US wildfire tail actually runs (~$28B of exposure from 2020 Labor Day fires). California CCAs have taken ~42% of what would have been PG&E’s bundled load (PowerMag). LADWP and the public-power camp are the political alternative — SF’s periodic municipalisation pushes are the recurring narrative pressure. Sunrun, Tesla Energy, Enphase, SunPower successors attack from behind the meter, with Sunrun already dispatching ~361 MW of California VPP into summer peak (Latitude Media, 2025). Base Power, Bloom Energy, Enchanted Rock attack C&I with fuel-cell/gas/battery hybrids that let large customers — especially data centres — soft-defect from the grid.

History and evolution

• 10 October 1905 — Pacific Gas and Electric incorporated from CG&E + SFG&E merger.
• 18 April 1906 — San Francisco earthquake; PG&E rebuilds.
• 1997 — HoldCo structure adopted.
• 6 April 2001 — First Chapter 11 during California electricity crisis.
• April 2004 — Emergence from first Chapter 11 after ~$10.2B in creditor payments.
• 9 September 2010 — San Bruno pipeline explosion; 8 killed.
• 2015 — $1.6B CPUC fine and ~$500M civil settlements on San Bruno.
• August 2016 — Federal 6-count conviction including obstruction of justice.
• October 2017 / November 2018 — Wine Country and Camp fires attributed to PG&E; 100+ killed.
• 29 January 2019 — Second Chapter 11.
• 12 July 2019 — Newsom signs AB 1054.
• 6 December 2019 — $13.5B Fire Victim settlement.
• 1 July 2020 — Emergence from second Chapter 11; Fire Victim Trust funded with $6.75B cash + 22.19% of stock.
• 4 January 2021 — Patti Poppe becomes CEO.
• July 2021 — 10,000-mile undergrounding pledge.
• 4 May 2023 — Carolyn Burke appointed CFO.
• December 2023 — FVT completes final divestment; dividend reinstated at $0.01/quarter.
• 31 October 2024 — Kerry Cooper succeeds Robert Flexon as Chair.
• 5 August 2025 — FERC TO21 base ROE approved at 10.38%.
• 23 October 2025 — 1,000th undergrounded mile energised.
• March 2026 — $24/month Base Services Charge effective.
• 24 July 2026 — Q2 2026: 12.7 GW pipeline; $73B five-year plan; guidance reaffirmed.
• 31 August 2026 — SB 492 dies without subrogation shield; PCG -20%.
• 1 September 2026 — Mizuho, BMO, Wells Fargo downgrades.
• 2 September 2026 — Strategic Review Committee launched; $2B 2027 capex deferral.

What people say

The case for. Poppe has done real work. The undergrounding programme hit its first 1,000-mile milestone in October 2025 (TD World; Plumas Sun). Safety metrics have improved on disclosed SIF tracking. Moody’s upgraded PG&E Corporation and Pacific Gas & Electric during 2024 (Investing.com). The dividend was reinstated in December 2023 for the first time since 2017. The 12.7 GW data-center pipeline (Energy Connects, July 2026) is among the largest visible in US utility sector. Rate-base compounding math is clean: $73B of capex through 2030 at a ~10.28-10.38% authorised ROE is arithmetic even a cautious analyst can model to 9%+ EPS CAGR.

The complaints. SB 492 exposed the fragility. Analyst downgrades on 31 August-1 September 2026 explicitly cited unresolved wildfire liability (Wells Fargo; BMO; Mizuho, per Yahoo Finance and Trading Pedia). Ratepayer-advocate voices — the CPUC’s own Public Advocates Office, CalMatters, KQED — flag that Californians pay roughly double the national average for electricity (34.74¢/kWh vs 18.34¢/kWh, Electric Choice September 2026), and that the data-center-lowers-bills argument is conditional on load-shape assumptions that may not hold. Wildfire-survivor coverage is corrosive: KQED, Press Democrat and KRCR reported in December 2023 that FVT survivors ended up materially short of the promised recovery because the Trust sold PG&E shares below the plan-confirmation assumption. The 2015-2018 wildfire cycle, the 2010 San Bruno explosion and the 2016 felony obstruction conviction are still the first three items in every wildfire-litigation opening statement. Credit context: PG&E is still low-investment-grade (roughly BBB-/Baa2/BBB-) and the September 2026 strategic review explicitly names investment-grade status as a target — meaning management does not treat the ratings as safe.

Outlook: well positioned or at risk?

At-risk. Fundamentals are not fragile in the pre-2019 sense — balance sheet, safety programme and rate-base plan are all materially healthier than at emergence — but three interlocking pressures now point the same direction and SB 492 is only the most visible.

First, cost of capital. Regulated utility equity is fundamentally a levered bond proxy plus a small growth-per-share adder; the discount-rate assumption does most of the work in any DCF. When the state’s own legislature declines to price out the wildfire tail, the market must, and the market’s estimate is not $21B — it is uncapped. The 20% single-day drop on 31 August 2026 is not a trading anomaly; it is the tail-risk premium moving from AB 1054’s implicit ceiling to no ceiling. Every dollar of PG&E’s $73B plan now needs to earn a higher return to justify the equity issuance required to fund it.

Second, cross-subsidy politics. California residential retail rates near $0.35/kWh are already politically unsustainable. Every incremental dollar of wildfire-mitigation, undergrounding and data-center interconnection flows to the same rate base recovered from the same customer bill. The March 2026 Base Services Charge is a first-order response, but the second-order response — behind-the-meter defection via rooftop solar + battery + VPP + microgrid — is what Sunrun’s 361 MW dispatch already demonstrates.

Third, structural erosion at both ends. CCAs took ~42% of bundled load (PowerMag). Data-center load may add gigawatts but concentrates operational risk on interconnection queues and CAISO transmission upgrades. Rooftop-plus-battery erodes residential fixed-cost recovery. AI demand is real, but its capture is contested by hyperscaler-owned generation, behind-the-meter fuel cells and merchant developers.

What flips this back to well-positioned: SB 492 or an equivalent shield in the 2027 session; a rating upgrade to solid investment grade; a data-center pipeline conversion rate that materially outperforms; a strategic-review outcome (e.g., a transmission carve-out) that unlocks the $2B 2027 deferral without diluting equity. Absent that, this is a compounding-machine story trading at bond-proxy multiples with an equity tail the market now believes is uncapped.

How to attack it

The obvious vector is behind-the-meter, distributed-and-orchestrated power for the residential and small-commercial customer that decouples the household from the volumetric grid charge. The arithmetic is decisive: PG&E’s ~39¢/kWh bundled residential rate (Solar.com, March 2026) versus a fully-installed rooftop-solar-plus-battery LCOE in the low-to-mid teens. Sunrun’s California VPP already dispatches 361 MW into summer peak. A well-capitalised attacker that packages rooftop + battery + smart-panel + VPP-revenue-share as a subscription — the Base Power / Sunrun Shift / Tesla Powerwall model on modern software — takes the arbitrage between PG&E’s volumetric rate and its own installed cost, keeps the VPP dispatch revenue, and offers the household outage resilience the incumbent structurally cannot match during PSPS events.

Second vector: wildfire-liability MGA and specialty reinsurance capital. SB 492’s collapse leaves the subrogation door open, meaning insurers will keep suing utilities to recover fire-payout losses. Capital that prices wildfire-liability catastrophe risk on modern models — parametric wildfire covers, utility-financial-guarantee wrappers, wildfire cat bonds — has real product-market fit. A purpose-built utility-wildfire MGA on modern paper is the founder-scale wedge; Berkshire, Aon, Guy Carpenter, RenaissanceRe and specialty ILS platforms are writing pieces of it already.

Third: data-center microgrids and behind-the-fence generation — the Base Power / Bloom Energy / Enchanted Rock playbook aimed at PG&E’s 12.7 GW queue. The queue itself is the bottleneck. Any hyperscaler that can bring 100-500 MW of on-site fuel-cell + gas + battery online in 18-24 months, rather than wait for a four-to-seven-year PG&E interconnection cycle, is a customer for a founder-built energy-as-a-service business.

Weaknesses to exploit: (1) the wildfire tail is uncapped post-SB 492; (2) rate design is 15 years behind the DER cost curve; (3) interconnection queues are structurally long; (4) undergrounding is 1,000 of 10,000 miles complete at October 2025; (5) CCAs continue to reduce captive-load footprint; (6) the balance sheet requires continual equity issuance to preserve the 52% equity ratio at $73B of capex; (7) 39¢/kWh residential rates make every rate-case ask a public fight; (8) legacy IT and OT limits software-defined grid capabilities; (9) the 2 September 2026 strategic review is a management-level admission the current structure is not optimal; (10) any large California wildfire in 2027-2028 attributed to PG&E equipment is a legitimate re-run of 2019 given the exhausted subrogation shield.

Adjacent-segment play

The obvious adjacent play is a carve-out or partial monetisation of PG&E’s regulated transmission business. Regulated transmission trades at richer multiples than integrated IOUs because earnings are formula-rate driven (FERC 10.38% base ROE on TO21), growth is unambiguously positive (CAISO transmission expansion for data-center load and renewables), and wildfire-liability profile is more contained than distribution. A partial IPO of PG&E Transmission — modelled on ITC Holdings pre-Fortis, or NextEra Energy Partners — could crystallise value for HoldCo shareholders and reduce equity needs at OpCo. The 2 September 2026 strategic review explicitly names “regulatory, financial, operational and strategic alternatives” and this is the most obvious lever.

A related play: a Bermuda-domiciled specialty reinsurance vehicle for AB 1054 exposure, capitalised by PG&E/SCE/SDG&E and third-party ILS capital that reinsures its shareholders’ wildfire tail — the marine-insurance P&I mutual-club template. It would not eliminate the tail, but would move it off the balance sheet into a market-priced instrument.

Third: data-center co-location and behind-the-fence generation as a separate infra vehicle. PG&E’s transmission rights-of-way, substation real estate and interconnection queue slots are strategic assets with hyperscaler demand at scale. A JV with Blackstone Infrastructure, KKR, Brookfield or ArcLight — the Sempra Infrastructure LNG carve-out template — could capture a share of hyperscaler capex without loading it onto PG&E’s equity base. Southern Company’s separate merchant Southern Power business and NextEra Energy Resources versus FP&L are the operating precedents.

Sources and further reading

• PG&E drops 20% on California wildfire bill — Seeking Alpha, 31 August 2026.
• PG&E Sinks 18%, Edison International Tumbles 23% as California Wildfire Bill Omits Liability Cap — 24/7 Wall St, 31 August 2026.
• PG&E downgraded by Wells Fargo, BMO as California wildfire reform falls short — Yahoo Finance, 1 September 2026.
• Deal on California Wildfire Liability Collapses in Legislature — KQED, September 2026.
• PG&E Launches Strategic Review to Build a Better, More Affordable Energy Future — PG&E investor relations, 2 September 2026.
• PG&E Q2 2026 Results — Core EPS Guidance Reaffirmed — StockTitan / PG&E, 24 July 2026.
• PG&E Corp Form 10-Q Q2 FY2026 — SEC EDGAR, August 2026.
• California AB 1054 (2019-2020) — California Legislative Information.
• PG&E Completes 1,000 Miles of Underground Powerlines — TD World, October 2025.
• PG&E Corporation Reinstates Common Stock Dividend — PR Newswire, December 2023.
• Fire Victim Trust sells final chunk of PG&E stock — Press Democrat, December 2023.
• PG&E Convicted of Obstructing Blast Probe, Breaking Safety Laws — SF Chronicle, August 2016.
• Going Way Back to the Last Time PG&E Declared Bankruptcy — Power Engineering.
• California Electricity Costs (September 2026) — Electric Choice, September 2026.
• Should Investor-Owned Utilities Be Worried About Community Choice Aggregation? — Power Magazine.