On May 11, Ford established a wholly-owned subsidiary, Ford Energy, and invested $20 billion to retrofit its Glendale plant in Kentucky, aiming to produce 20 GWh of energy storage systems per year. The next day, Morgan Stanley analyst Andrew Percoco separately valued this business at $100 billion. The market reacted swiftly, with Ford's stock price rising from $11.99 to $14.48 over two trading days, adding approximately $8 billion to its market value.
The story seems to follow the standard script of "traditional auto companies being revalued by AI." However, a closer look at the supply-side ledger reveals another name: Louisiana.
What Is the Market Reassessing?
Morgan Stanley's $100 billion valuation of Ford Energy was not arbitrary. Percoco's report assumed that Ford Energy would achieve a 25% gross margin and $346 million EBIT in 2028, and then applied the market-to-sales ratio of leading energy storage companies in reverse. Compared to Ford's existing business profit structure, this is a completely different species.
Ford's 2024 financial report shows that Ford's Blue Fuel division had annual revenue of $145.4 billion, EBIT of only $5.28 billion, and a profit margin of 3.6%. The gross margin for selling one F-150 is estimated in the industry to be between 10-13%. In the same period, the energy storage industry is in a completely different league. According to various companies' financial reports, Tesla's Megapack business has a gross margin of 26.2%, and SunPower's energy storage business exceeded 28% in the first half of 2024. Pure system integrator Fluence ranges from 12-14%. Morgan Stanley is betting that Ford's in-house LFP cell addition and the cost structure of retrofitting existing EV factories can fall into the top tier, with a gross margin per GWh at least double that of selling one F-150.
This is the reason why the market is willing to pay a premium. However, the market has only given $80 billion so far, and Morgan Stanley's $100 billion target implies that there is still $20 billion yet to be absorbed. The reassessment is still in its early stages.
To absorb this $20 billion, Ford Energy needs to realize its production capacity, contracts, and pricing. Production capacity is set to be available for the first batch delivery in the second half of 2027, while contracts will depend on downstream buyers, the four most competitive regions in electricity procurement.
The Four Corridors
In the same research report, Morgan Stanley provided another figure. The installed power capacity of U.S. data centers was 40 GW in 2024, projected to reach 79 GW by 2027, with an approximately 49 GW gap between demand and supply in 2028, equivalent to a 20% shortfall. This gap will not be evenly distributed across all 50 states.
Based on currently publicized data center power requests, the gap is concentrated in four corridors. In Arizona, the APS service area in Phoenix has a pending queue of 30 GW, Loudoun County in Virginia has signed or nearly signed capacity of 14.2 GW, in Texas, Abilene alone has the Stargate project led by OpenAI with 1.2 GW, and Hillsboro in Oregon has an additional 0.4 GW. The two sets of figures have different calibers: the 30 GW in Phoenix includes projects in the utility queue that have not yet been finalized, while the 14.2 GW in Loudoun represents the actual capacity close to signing.
This is the demand-side foundation of Ford's $10 billion valuation. When a region's grid capacity growth cannot keep up with the speed at which data centers are grabbing power, energy storage arbitrage opportunities are created. On one side is the training load that is almost impossible to cut during peak hours, and on the other side is the supply curve that the grid cannot respond to immediately. The space left for energy storage companies is a deterministic cash flow.
But taking power comes at a cost. The question is, who pays.
The Bill Lands in Louisiana
On May 14, Fortune's front page reported Meta's Hyperion project in Richland Parish, Louisiana. This is Meta's flagship data center betting on "AI Advantage," with a cost of $10 billion. Approved by the Richland Parish Council in July 2024, a total of $3.3 billion in sales and use tax exemptions will be covered over 20 years.
This exemption exceeds the total sum of the Louisiana Police Department's budget for 7 years.
Meta also obtained a 60% local property tax exemption through a PILOT agreement, contingent on creating 300 local permanent jobs. The electricity consumption forecast for this project may account for 20% of the state's total electricity consumption.
A more critical comparison lies across state dimensions. Sales tax exemptions for data centers in Texas for the 2025 fiscal year amount to approximately $1 billion, making it the most expensive industry subsidy in Texas. For Virginia in the 2024 fiscal year, it is $1.02 billion, and for Illinois in 2023, it is $371 million. The $3.3 billion subsidy obtained by Meta for a single project surpasses the total amount of all data center subsidies in any state for an entire year.
This is not an isolated case. Good Jobs First estimates that at least 36 states have enacted special legislation to authorize sales tax exemptions for data centers, with only 11 states publicly disclosing the list of beneficiary companies. Some states' official calculations are more direct, showing that for every $1 in data center tax breaks, the state incurs a net loss of 52 to 70 cents. Ford's soaring $80 billion valuation is built on the implicit assumption that "this subsidy arms race will continue."
The subsidy arms race relies on state legislators being able to explain why a tax break exceeding a police budget spanning over 7 years is necessary. This is becoming increasingly challenging.
Data Center Opposition Rate Surpasses Nuclear Power Plants
A Gallup nationwide survey in 2025 showed that 70% of respondents opposed having a data center built near their homes. In the same survey, the opposition rate for nuclear power plants was 53%. Data centers now have the highest infrastructure opposition rate in the U.S.
This is not just a statistical figure. As of early 2026, Data Center Watch's database shows that 69 jurisdictions have passed moratorium resolutions or local referendums to reject data center approvals, leading to $64 billion worth of planned projects being put on hold. In 2024, Loudoun County, Virginia, lost $267 million to the public school system due to data center tax breaks, and this issue has now captured the voters' attention.
The reasons behind the annoyance caused by data centers are more complex than those of nuclear power plants. Nuclear power plants are usually located dozens of kilometers away from residential areas, while data centers are often situated close to urban areas. Data centers consume coolant water, increase the strain on the power grid, and generate low-frequency noise, but the local employment they provide is mostly in the range of hundreds, significantly lower than the job density in steel or automobile factories. Hyperion's 326 operational positions correspond to $3.3 billion in tax breaks, equivalent to a fiscal cost of $10 million per position. The room for local government explanations is shrinking.
Ford Energy's $100 billion valuation's realization path depends on the continuous operation of this power grid, the maintenance of subsidies, uninterrupted power supply, and voices of opposition not turning into legal mandates. If any part of this chain loosens, reverse pricing will follow suit.
Ford's $80 billion surge, Meta's $33 billion subsidy, Phoenix's 30 GW queue, the 70% opposition rate – these are readings from four snapshots of the same power grid. They are still loose data points to this day. Once one hard constraint emerges, whether it's a physical power shortage, a state financial breach, or a public referendum, the others will simultaneously undergo a repricing.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
