By The Sooner Sentinel
Investigative Article
August 31, 2026 — Oklahoma politicians moved hundreds of millions of dollars toward America’s largest proposed aluminum smelter before the public had the completed environmental study, the final power arrangement, or a detailed spent-potliner management plan. The question isn’t whether Oklahoma should manufacture things. It is whether taxpayers and neighboring landowners are being asked to assume risks that private investors should bear themselves.
“The purpose of life is to find a mode of being that is so meaningful that the fact that life is suffering is no longer relevant,” Dr. Jordan Peterson once observed, and few things make that suffering more needless than the petty tyrannies of local bureaucrats who mistake regulation for wisdom.
There is another kind of government failure, however, that advocates of limited government should recognize just as quickly. It happens when government stops merely regulating private enterprise and begins choosing, subsidizing and financially engineering it. Oklahoma’s proposed aluminum smelter at the Tulsa Port of Inola deserves scrutiny from precisely that perspective.
The project is enormous. Oklahoma Primary Aluminum, a joint venture owned 60 percent by Emirates Global Aluminium and 40 percent by Chicago-based Century Aluminum, proposes producing 750,000 metric tons of primary aluminum annually, making Inola the largest primary aluminum plant ever constructed in the United States. The developers project approximately 1,000 permanent jobs.
There is a serious national-security argument for rebuilding domestic aluminum production. America imports a large portion of the primary aluminum it consumes, while aluminum remains essential to aerospace, automobiles, construction, electrical infrastructure and defense. That argument should not be dismissed merely because the project is controversial.
But neither should “national security” become a magic phrase that exempts a $4-billion industrial project from questions ordinary Oklahomans would be expected to answer before building a barn.
And some rather large questions remain unanswered.
Start With the Number Nobody Can Ignore: 1,000 Megawatts
Oklahoma has spent much of the past two years arguing about data centers.
Attorney General Gentner Drummond’s lawsuit describes the proposed smelter as requiring more than 1,000 MW continuously. The aluminum industry’s own national trade association says a single new aluminum smelter can consume approximately 11 terawatt-hours of electricity every year, comparable to the annual electricity consumption of a major American city such as Nashville or Boston.
At 1,000 MW, Inola would represent:
13⅓ times Oklahoma’s 75-MW statutory threshold for a large data center.
Compared with a 20-MW facility, the smelter is approximately 50 times larger. Compared with a 10-MW data center, it is approximately 100 times larger. Even a massive 100-MW computing facility represents only about one-tenth of a 1,000-MW smelter. And unlike many computing workloads, aluminum electrolysis is extraordinarily difficult to interrupt.
Inside an aluminum smelter, electricity isn’t simply running computers, lights or motors. It maintains the electrochemical process inside hundreds of reduction cells, or “pots”, containing molten electrolyte. A sufficiently prolonged power interruption can allow pots to freeze, producing enormous restart costs and potentially damaging equipment.
This isn’t an industrial customer that casually turns itself off during a hot August afternoon. It wants enormous amounts of electricity 24 hours a day, 365 days a year.
That makes the Legislature’s treatment of data centers particularly interesting, more on that late in the article.
Who Pays for the Gigawatt?
Public Service Company of Oklahoma says the smelter will pay the costs associated with the electric service it receives.
That is important.
But PSO has also acknowledged that investments, costs and revenues associated with serving the project ultimately will be examined by the Oklahoma Corporation Commission in future rate proceedings. The precise infrastructure and service arrangements were still being developed this summer.
Meanwhile, PSO has launched a remarkable procurement.
Its 2026 All-Source Request for Proposals seeks up to 4,000 MW of accredited generation capacity, with resources expected in Oklahoma and commercially operational by November 30, 2029. PSO says those resources may include natural gas, wind, solar, batteries, hydrogen and other technologies.
Four thousand megawatts. The proposed Inola load alone could equal roughly one-quarter of that number. That does not establish that one-quarter of PSO’s procurement is being built for Inola. Load forecasting is more complicated than that, and PSO serves other growth. But the coincidence of scale and timing demands transparency.
The smelter is expected around 2030. PSO’s new resources are targeted for 2029. And Oklahoma Commerce confirmed in January that negotiations with PSO for a competitive long-term power supply were progressing. The public deserves to see the resulting arrangement. Not a press release. We need to see the contract.
Then There Is the $255 Million Electricity Arrangement
This story gets considerably more interesting. Oklahoma Primary Aluminum says the project’s $255 million state incentive does not simply arrive as a check written to EGA or Century.
According to the company’s own explanation, the state treasurer places the $255 million LEAD package into an interest-bearing investment fund. Once operations begin, equal annual payments are to be made to PSO for 15 years, with those payments credited against the smelter’s electric bill.
Read that again.
Oklahoma taxpayers are putting $255 million into a structure whose eventual payments reduce the project’s electricity bill.
Whatever terminology government accountants attach to that arrangement, taxpayers are entitled to ask whether government is effectively helping make an enormous industrial electric load economically viable.
The project has also been selected for up to $500 million in federal support, according to state and federal-project reporting.
And Oklahoma lawmakers created the Reindustrialize Oklahoma Act through HB 2781, providing performance-based incentives for qualifying manufacturers investing at least $2 billion and meeting substantial job requirements.
This produces a peculiar “version” of free enterprise. Government subsidizes the industrial development. Government participates in lowering its electricity expense. Government oversees the monopoly utility serving it. Government regulates the emissions. Government issues the permits.
And if something goes wrong decades later, taxpayers will reasonably want to know who remains financially responsible. That isn’t laissez-faire capitalism. It is government-managed industrial policy. Perhaps policymakers can justify it on national-security grounds. But they should call it what it is.
The Political Irony
Oklahoma’s debate over data centers offers an almost perfect comparison.
In 2026, the Legislature enacted the Data Center Customer Ratepayer Protection Act, drawing a bright statutory line at 75 megawatts. A new data center, AI-computing facility or cryptocurrency operation adding 75 MW or more is classified as a “large load customer,” and the law embraces a straightforward principle known as cost causation: or, customers responsible for causing new electric-system costs should bear their equitable share of those costs. But there is an extraordinary exception buried in that definition: traditional industrial ratepayers are specifically excluded.
Here is why that matters in Inola. The proposed aluminum smelter is expected to require more than 1,000 MW of continuous electricity, more than thirteen times Oklahoma’s 75-MW threshold for a covered data-center load. Yet because an aluminum smelter is an industrial customer rather than a data center, it does not fall within the statute’s definition of a “large load customer.”
Data centers certainly deserve scrutiny. They can create enormous new electric loads requiring additional generation and transmission infrastructure. Oklahoma lawmakers were right to ask who pays when those costs are created. But intellectual consistency requires asking precisely the same question about Inola.
If a 75-MW data-center load deserves special statutory protection to keep infrastructure costs from being shifted onto Oklahoma families, farmers and small businesses, what level of scrutiny should accompany an industrial facility demanding more than 1,000 MW around the clock?
The electrons don’t care whether they ultimately power servers or aluminum reduction pots. Neither does the transmission line. Neither does the generating plant. And neither does the Oklahoma ratepayer opening their next electric bill.
The question isn’t whether Oklahoma should welcome industry. It should. The question is whether Oklahoma will apply the same limited-government principle to every politically favored industry: Those who create the cost should pay the cost.
That, after all, is the very principle the Oklahoma Legislature wrote into law in 2026.
The Second Gigantic Question Is Beneath Our Feet
Electricity is visible on a spreadsheet. Groundwater is not. Oklahoma should understand that distinction better than almost any state in America.
Travel west to the Panhandle and you can see what happens when groundwater withdrawals become a multigenerational economic assumption. The Ogallala aquifer isn’t an environmentalist abstraction. It is infrastructure deposited by geology. And we have been spending it.
A new 2026 USGS study of Oklahoma’s Panhandle and northwestern Ogallala found average groundwater withdrawals during 1998–2022 of approximately 422,054 acre-feet annually in the Panhandle portion, while estimated recharge was only about 175,068 acre-feet annually. Recharge averaged just 0.63 inch per year.
In other words, in that portion of Oklahoma, withdrawals were roughly 2.4 times modeled recharge during the study period.
That is the lesson Inola should learn from western Oklahoma. Not that northeastern Oklahoma is presently running out of groundwater. It isn’t. The lesson is that water policy decisions made when an aquifer appears abundant can become somebody else’s crisis 40 years later.
Oklahoma’s “Maximum Annual Yield” Isn’t Necessarily Sustainability
This is one of the least understood features of Oklahoma water law. The Oklahoma Water Resources Board establishes a Maximum Annual Yield, or MAY, for groundwater basins. The phrase sounds reassuring. But MAY should not be confused with a guarantee that withdrawals equal natural recharge forever.
Oklahoma’s groundwater-management framework evaluates how much groundwater can be allocated under statutory basin-life criteria. Landowners then receive an Equal Proportionate Share, generally based on acreage.
That can permit deliberate depletion over time. The Ogallala demonstrates the difference between managing depletion and preventing depletion. And northeastern Oklahoma now has an unusual opportunity not to repeat that history.
The Aquifers Under and Around Inola
The important regional bedrock aquifers are the Boone and Roubidoux, also known respectively as the Springfield Plateau and Ozark aquifers. USGS completed a major groundwater-flow study in cooperation with OWRB in 2024.
And here is the remarkable part: OWRB had not yet legally established final Maximum Annual Yields for either aquifer when the USGS report was published.
The study modeled potential Equal Proportionate Share withdrawal rates.
For the Boone: 1.10 acre-feet/acre/year — 20-year scenario
0.98 — 40-year
0.96 — 50-year
For the Roubidoux: 1.76 acre-feet/acre/year — 20-year
1.34 — 40-year
1.25 — 50-year.
The modeled decline as the planning horizon lengthens illustrates the point beautifully: how much water appears “available” depends partly upon how long you expect the resource to last. Current withdrawals are relatively modest, and USGS’s modeling does not portray Boone-Roubidoux as another Ogallala today.
That’s good news. But abundance today isn’t an argument for indifference tomorrow. It is an argument for establishing the baseline before transformative industrial demand arrives.
For Inola, Water Quantity May Not Be the Biggest Water Question
There is an important distinction here. Based on currently available information, it would be premature to claim that this smelter will drain the Boone or Roubidoux aquifers.
The project’s final water balance, groundwater withdrawals and surface-water requirements have not been publicly developed enough to support that conclusion.
The more immediate concern may instead be water quality. That brings us to two words every resident near a primary aluminum smelter ought to learn: Spent potliner.
What Comes Out of 484 Aluminum Pots?
The proposed plant uses electrolysis to turn refined alumina into primary aluminum. That distinction eliminates one frequently repeated concern. Inola is not a bauxite refinery. It therefore should not generate the enormous alkaline “red mud” waste stream associated with the Bayer refining process.
That criticism should be, at least temporally, removed from serious discussion of this project. But primary aluminum smelting produces another troublesome waste. The carbon and refractory lining inside an aluminum reduction pot eventually deteriorates and must be removed. That material becomes spent potliner—SPL.
EPA identifies spent potliners from primary aluminum reduction as RCRA hazardous waste K088. And EPA’s historical record explains why. Spent potliner can contain significant concentrations of cyanide and fluoride, along with other constituents. In establishing treatment requirements, EPA documented actual cases of groundwater contamination associated with improperly managed potliner.
At the Kaiser Aluminum Mead Works Superfund site in Washington, EPA concluded that spent potliner and pot-soaking liquor were primary sources of cyanide and fluoride contamination in soil and an on- and off-site groundwater plume. EPA reported spent-potliner fluoride averaging approximately 80,600 mg/kg in samples cited in its remedial record.
That doesn’t mean Inola will become the next Mead Works. Modern waste-management technology, hazardous-waste regulations and engineering standards are vastly better than the practices that created many legacy contamination sites. But history establishes something important: The hazard is real even if the probability of release can be engineered downward.
At 750,000 Tons a Year, “Temporary” Becomes an Important Word
Industry estimates vary with technology and pot life, so nobody should pretend we know Inola’s exact annual SPL production before the final design and waste plan are public.
Using the user’s supplied planning range of roughly 20–40 kilograms of SPL per metric ton of aluminum, a 750,000-tonne smelter could eventually generate on the order of: 15,000–30,000 metric tons of spent potliner per year once pot relining reaches steady-state replacement cycles.
Using the rough midpoint assumption of one ton per 40 tons of aluminum gives approximately: 18,750 tons per year. These are planning estimates, not a disclosed Inola waste forecast. That distinction matters. Because the company has not publicly provided one.
The Missing Bunker Numbers
Oklahoma Primary Aluminum says waste materials will be temporarily held in “sealed concrete bunkers” before being shipped off-site for recycling or processing. The company also promises baseline and continuing groundwater and soil monitoring.
Those are sensible commitments. But as of this writing, publicly available project material does not answer several elementary engineering questions:
How many bunkers? How large? How many tons of K088 waste can they contain? How many months, or years, of SPL generation does that represent? What happens if the contracted recycler stops accepting material? What is the secondary containment design? Where are the bunkers relative to drainage paths and the river? What is the design-basis flood? How will contaminated firewater be contained? What groundwater monitoring wells will be installed downgradient? Who posts financial assurance if the joint venture someday closes?
These aren’t arguments against aluminum. They’re questions one asks before storing reactive hazardous waste beside a river. The company’s voluntary Environmental and Social Impact Assessment doesn’t yet provide those answers.
Its ESIA webpage says waste-management commitments will eventually be established in the Environmental and Social Management Plan. Emergency-response, spill-prevention and weather-resilience plans are likewise still being developed.
That is the fundamental problem with much of today’s public debate. The government has already committed extraordinary political and financial support. The engineering answers are still arriving.
And the Verdigris River Is Right There
This isn’t an industrial facility sitting miles from surface water. The project’s own March 2026 ESIA scoping summary describes a 437.5-acre site approximately three miles west of Inola with direct access to the Verdigris River.
That river access is one of the site’s great economic advantages. Alumina and other bulk commodities can move by barge through the McClellan-Kerr Arkansas River Navigation System. But geography doesn’t change when the barge leaves. The river remains there during thunderstorms, floods, equipment failures, fires and human mistakes.
The potential pathways deserve analysis:
stormwater → river;
leachate → shallow groundwater;
groundwater → streams and river;
treated wastewater → receiving waters;
airborne fluoride deposition → soil and surface runoff.
None of those arrows proves contamination will occur. They identify pathways that the final engineering must convincingly interrupt.
“Sealed Concrete Bunkers” Is a Beginning, Not an Answer
Concrete is useful. Concrete is not magic. Concrete cracks. Joints fail. Coatings deteriorate. Drainage systems clog. Forklifts hit things. Storms happen. Contractors make mistakes. Companies change ownership. Waste recyclers fail. Plants close.
The appropriate response isn’t another mountain of bureaucratic paperwork. It is financial accountability. A limited-government solution should start with a simple principle: The people creating a risk should bear its full cost.
That means stringent financial assurance from the project owners sufficient to cover closure, hazardous-waste removal and credible groundwater remediation—not a future appeal to Oklahoma taxpayers if the economics look different in 2055.
Government’s legitimate role isn’t to micromanage every bolt. It is to protect property rights and prevent one person’s industrial activity from imposing uncompensated costs on somebody else’s farm, well or river.
The Project Is Not Environmentally Approved Yet
This point has sometimes been lost amid the political rhetoric. As of late August, Oklahoma DEQ says it is reviewing the project’s air-quality construction permit application.
DEQ’s project page still lists the Best Available Control Technology technical meeting, dispersion-modeling technical meeting and draft air-quality permit meeting as TBD. DEQ also notes that a facility of this kind can require separate permits involving air emissions, water discharge, stormwater and waste handling.
The voluntary ESIA is also not the equivalent of a completed regulatory approval. The project’s own ESIA page says the assessment is being conducted to inform an eventual Environmental and Social Management Plan.
Meanwhile, the legal fight has intensified. Attorney General Drummond sued in June to block the project, alleging threats to residents, livestock and natural resources. The developers dispute those allegations and say they will meet stringent environmental standards.
On August 11, Drummond sought a preliminary injunction against construction, and litigation has also erupted over whether the case belongs in state or federal court. So this is not a permitted plant whose environmental performance has already been demonstrated. It remains a proposed plant undergoing permitting, environmental analysis and litigation.
But Don’t Blame the Data Center for the Surveillance State
There is another distinction worth making.
Oklahomans concerned about automatic license-plate readers, facial recognition, mass surveillance and government databases have legitimate civil-liberties questions.
But a data center is infrastructure. It stores and processes information. The same infrastructure backs up family photographs, processes bank transactions, provides maps, supports hospitals, runs businesses and makes cloud computing possible.
The problem isn’t the building full of servers. The problem is who collects the information, under what authority, for what purpose, how long it is retained, who receives it, and whether government must obtain lawful process before searching it.
Blaming data centers for ALPR surveillance would be a little like blaming aluminum for an unconstitutional police search because patrol cars contain aluminum.
Limited government requires identifying the actual government power being abused.
Likewise, Aluminum Isn’t the Enemy
Aluminum matters. A country that cannot produce basic industrial materials eventually becomes dependent upon countries that can. Aircraft need it. Electrical systems need it. Vehicles need it. Defense manufacturers need it.
Rebuilding American primary aluminum production is a legitimate strategic objective.
The question is whether rebuilding American industry requires Oklahoma taxpayers to subsidize a particular multinational joint venture while assuming poorly quantified infrastructure and environmental risks.
Those are different propositions. Free enterprise means investors receive the reward when they are right. It also means investors, not captive ratepayers and taxpayers, should absorb the consequences when they are wrong.
The Ogallala’s Warning to Inola
The Ogallala isn’t an argument that Inola’s aquifers are about to disappear.
It is a warning about institutional memory. Every generation believes its assumptions are reasonable. The water appears plentiful. The economic development appears necessary. The models say there is capacity. The permits say withdrawals are lawful.
The next project is only one more project. Then decades pass.
USGS now estimates that Oklahoma Panhandle Ogallala withdrawals during its 1998–2022 study period substantially exceeded modeled recharge. Northeastern Oklahoma has an advantage western Oklahoma no longer possesses: Time.
The Boone and Roubidoux aquifers aren’t presently experiencing the same depletion crisis. So establish defensible Maximum Annual Yields. Establish groundwater chemistry before the first pot is energized.
Map shallow groundwater flow toward the Verdigris. Establish private-well baselines. Publish the monitoring locations. Determine the smelter’s actual water balance. Determine the SPL generation rate. Publish the bunker capacity. Identify the recycler. Publish the transportation route. Require financial assurance. Then let everyone see the numbers.
That isn’t hostility toward industry. That’s what property rights look like when the property extends underground.
The Questions Oklahoma Officials Should Answer Before Construction
Before taxpayers, ratepayers and Inola residents are told that this project is effectively a done deal, state officials and Oklahoma Primary Aluminum should put a single public record together answering the questions that remain scattered among agencies, contracts and unfinished studies:
What is the final maximum and expected continuous electrical demand—in MW?
What is the projected annual consumption—in TWh?
What new generation and transmission facilities are required specifically because of the smelter?
Who pays every dollar of those facilities if the smelter is delayed, downsized or abandoned?
What does the final PSO power contract require, and what protections prevent cost shifting to other customers?
Why should an approximately 1,000-MW industrial customer receive less statutory ratepayer scrutiny than a 75-MW data center?
What is the plant’s complete annual water balance?
How much comes from the Verdigris, municipal systems and groundwater respectively?
What are the pre-construction water levels and chemical baselines in the Boone, Roubidoux and shallow groundwater?
How much spent potliner will the facility generate annually at full production?
What is the tonnage capacity of the sealed SPL bunkers?
How many days or months of production can they accommodate?
What happens if the off-site recycling contractor stops accepting K088 waste?
Where is the SPL storage area relative to the Verdigris, flood elevations and groundwater-flow direction?
What financial assurance guarantees cleanup if contamination appears 10, 20 or 40 years from now?
Those questions don’t require another government program. They require answers.
Just In
As Reported August 30, 2026 by Oklahoma Representative Tom Gann
Oklahoma Rep. Tom Gann has just recently added another layer to the growing questions surrounding the proposed Inola aluminum smelter, this time focusing not simply on environmental risk, electric rates or taxpayer incentives, but on who ultimately owns, supplies, finances, markets and benefits from a project being promoted in the name of American national security. Gann points to Treasury Secretary Scott Bessent’s newly announced Operation Economic Outcast, which is targeting the international corporate, financial and commodity-trading networks Iran uses to evade sanctions and generate revenue. Gann appropriately notes that Treasury has not accused or sanctioned Emirates Global Aluminium, Century Aluminum, Oklahoma Primary Aluminum or the UAE government. His argument is instead straightforward: if Washington believes national security requires looking beyond corporate nameplates and examining beneficial ownership, counterparties and financial relationships, Oklahoma should exercise the same diligence before committing substantial public resources to Inola.
The corporate relationships deserve that scrutiny. Oklahoma Primary Aluminum is proposed as a 60% EGA–40% Century Aluminum joint venture, while Glencore beneficially owns approximately 30% of Century, giving it roughly a 12% look-through economic exposure to the proposed venture, though not a direct 12% ownership interest in Oklahoma Primary Aluminum. More important than that arithmetic may be Glencore’s commercial relationship with Century. Century reported that approximately 54% of its 2025 consolidated sales went to Glencore. SEC filings also show Glencore purchasing Century aluminum, supplying alumina and other raw materials, participating in financial arrangements with Century and possessing a contractual right to designate a nominee to Century’s board. Century’s disclosures identify a Glencore alumina supply agreement for approximately 500,000 metric tons annually through 2028. None of that establishes that Glencore will supply, finance or market production from Inola, but it makes Gann’s demand that Oklahoma determine whether it will entirely reasonable.
Gann also points to a particularly sensitive disclosure buried not in political rhetoric, but in Century’s own SEC filings. Century reported that non-U.S. affiliates of Glencore had agricultural-product contracts involving Iranian entities wholly or majority owned by the Government of Iran. Century stated that the transactions complied with applicable sanctions requirements and emphasized that Century and its subsidiaries were not involved. Its 2025 Form 10-K nevertheless reported up to approximately $252 million in gross revenue associated with those contracts and said the Glencore affiliates expected to continue similar activities in compliance with applicable sanctions. Century’s subsequent first- and second-quarter 2026 filings do not establish that those transactions continued into 2026. Likewise, Glencore entities’ 2022 guilty pleas involving foreign bribery and commodity-market manipulation are historical facts deserving disclosure, but they do not constitute evidence of misconduct involving the proposed Oklahoma smelter.
For Oklahoma taxpayers, therefore, the issue is not guilt by association. It is due diligence before subsidy. If state officials are going to invoke national security while supporting a project involving a $255 million state financial structure, enormous electric demand and majority foreign ownership, taxpayers have every right to know whether Glencore will supply Inola’s alumina, purchase or market its aluminum, provide financing or hedging, influence Century’s decisions concerning the venture, or participate in moving its finished metal to customers. They should also know who those ultimate customers will be and what sanctions-compliance and end-user safeguards will apply. As Gann correctly frames the issue, these questions are not accusations, they are basic due diligence. Limited government should never mean government handing out public benefits first and asking difficult questions later. When taxpayers are being asked to share the financial risk, transparency should not be optional.
Gann’s article: Oklahoma Taxpayers Deserve Public Answers on the Inola Smelter
Post Script
All of this is particularly important for the farmers, ranchers and property owners surrounding Inola. If the project succeeds exactly as promised, its supporters will have an enormous industrial achievement to point toward. But if fluoride deposition damages forage, if a containment failure allows spent-potliner constituents to migrate into shallow groundwater, or if the project’s enormous electrical demand contributes to costs ultimately borne by ordinary ratepayers, the politicians who approved the project will not be the ones replacing a contaminated well, losing a hay crop or opening the monthly electric bill.
That is where Thomas Sowell’s warnings comes into play . . .
And yes, excellent piece.
So, where are we on getting a full EIS on the smelter? Even the infamous WPPSS projects underwent a ten year full EIS. Nothing should go forward without that. As for the concrete containment, I suggest they look into the casks employed by the nuclear industry to contain spent nuclear fuel rods.