For the first time since 1977, someone is building a new crude oil refinery in the United States.
America First Refining broke ground this month on a 168,000 barrel-per-day facility at the Port of Brownsville, Texas. Fluor is running front-end engineering. Axens is licensing the hydroprocessing, catalytic reforming, and isomerization units. Reliance Industries is backing it with capital on a scale that suggests this isn’t a one-off — it’s a template.
Three weeks later, the Strait of Hormuz is still closed. Eleven million barrels per day of Middle East crude production is offline. The 3-2-1 crack spread on the U.S. Gulf Coast is $54 per barrel — the highest it has been in a decade. Every existing U.S. refinery is running at 92% utilization and clearing margins that make compliance investments, reliability upgrades, and capacity restarts look cheap.
These two facts are not a coincidence. They are the same story told at different scales.
The Thirty-Year Drought Was a Policy Artifact, Not an Economic One
For three decades, the conventional wisdom was that American refining had peaked. Permitting was impossible. Environmental opposition was insurmountable. The economics of greenfield construction couldn’t beat incremental expansions at existing sites.
None of that was strictly true. What was true: domestic fuel demand had flattened, crack spreads were structurally narrow, and the cost of new RAGAGEP-compliant construction was genuinely punishing relative to bolting another unit onto an existing permit envelope.
The Hormuz crisis didn’t change the cost of construction. It changed the value of a barrel of distillate delivered from a secure domestic source. When the math changes by a factor of three in ninety days, a lot of previously unworkable projects become workable.
What This Means for Working Engineers
The next five years will demand a scale of domestic engineering capacity that doesn’t currently exist. Consider what a single new 168K BPD refinery requires:
A complete PSM program under 29 CFR 1910.119 — fourteen elements, from PHA to emergency response to mechanical integrity. An engineering standards library compliant with current RAGAGEP across ASME Section VIII, API 510/570/653, ASME B31.3, API 579, and dozens more. Operating procedures for every licensed unit. A mechanical integrity program with inspection intervals, fitness-for-service screening, and documented competency for every inspector. Management of Change infrastructure. Training programs for operators and maintenance craft. And that’s the pre-startup package — before the first barrel moves.
Multiply that by however many micro-refineries, modular units, renewable diesel expansions, and restart projects the next Hormuz-adjacent crisis triggers.
There are not enough PSM consultants in the country to deliver this the traditional way.
The AI Wedge
This is the moment AI in engineering stops being a conference slide and becomes load-bearing infrastructure. Standards crosswalks that used to take three to four weeks of senior engineer time can be done in three to five days when an AI tool is trained on the current edition library. PHA documentation that sat in Word templates can be generated from structured facility data and verified by an engineer instead of typed by one. Point cloud scans from a site walk can be converted to a P&ID starting point in hours rather than weeks.
The engineers who win the next decade are not the ones who resist these tools. They are the ones who use them to cover three greenfield projects simultaneously while their competition is still staffing one.
What I’m Betting On
I’ve spent most of this year building NEXUS CAD for lightweight 3D modeling and point cloud review, NORMEX Standards AI for engineering standards modernization, and NEXUS Compliance AI for PSM program management. The hypothesis was simple: the next refinery build cycle in the U.S. would come faster than most people expected, and the toolchain wasn’t ready.
AFR Brownsville moved that hypothesis out of theory. Hormuz moved it out of theory for every existing refinery on the Gulf Coast at the same time. The window to build the tooling these projects will actually use is now — not when the purchase orders are issued.
The fifty-year gap didn’t close because of a policy change. It closed because the geography of energy security finally showed up on the balance sheet. What engineers do with the next five years will determine whether domestic refining capacity is rebuilt as a twentieth-century industry or a twenty-first-century one.
Timothy Porritt is founder of Porritt Inc., building AI-powered tools for heavy industry including NEXUS CAD and NORMEX Standards AI. A petroleum engineer by training, Timothy writes about the intersection of industrial engineering, AI, and entrepreneurship.