QUESTION
This article is the second installment of
The Day of AI Reckoning, first published
on September 9, 2026, which is a response to a reader who
asked, in part:
A day of reckoning is coming, I promise you. Where's the regulation? Who's liable if AI makes a mistake? Who will reimburse me if I get sued because of an AI error? Why is there no quality control? Why are all these FinTech companies springing up all over the place with AI promotions?
Let's get real. I have no confidence that AI technology is a new type of economic revolution. Maybe it is for the billionaires who run it, since it seems to be a revolution in how fast they can become mega-billionaires.
Will there be a day of AI reckoning in mortgage banking?
COMPLIANCE SOLUTIONS
AI POLICY PROGRAM FOR MORTGAGE BANKING™
Policies and procedures for artificial intelligence related to mortgage banking compliance administration, mortgage loan origination compliance, mortgage servicing compliance, and industry best practices—including, but not limited to, applicable federal and state banking and consumer lending laws.
AI FOR MORTGAGE LOAN
ORIGINATION - SALES MANUAL
Sales Manual provides 9 chapters
covering the full AI-powered origination funnel in a 90-day implementation
roadmap, with clear team roles for fair lending, FCRA, and advertising
compliance guidance built into every chapter. Sample AI prompts, an ROI
tracking framework, and a compliance glossary.
RESPONSE
In Part II, we discuss:
- Capital Markets Are Starting to Ask the Same Question
- Public Is Rejecting the Infrastructure, Not Just the Valuations
- Data Center Opposition
____________________________________
Distributed after publication of all three articles.
Orders distributed first come, first served.
____________________
Here are a few recent related articles:
Capital Markets Are Starting to Ask the Same Question
Your skepticism showing up in mortgage compliance departments has a mirror in the capital markets. Through mid-to-late 2026, AI-adjacent equities have undergone a sustained, and at times sharp, repricing. Chipmakers lost over a trillion dollars in combined market value in a single stretch this summer amid concerns that AI infrastructure spending may be peaking faster than expected.
Software stocks have fared worse than chips in some cases. For example, as I write, Oracle is down more than 50% from its late-2025 high, Microsoft is down nearly 10%, and Meta is down more than 20%. Two contradictory investor theories are driving them: on the one hand, AI will make incumbent software obsolete; on the other, software companies have overspent on AI without adequate returns. One thing I learned at Wharton is that contradictory investor theories are a huge warning sign. And in this case, both narratives can't be true. The market's inability to settle on one signals unresolved uncertainty about what AI is actually worth in production, not just in a demo.
Meanwhile, the AI champs are rising. Whether you look at private-market valuations or explosive financial metrics, AI purveyors have grown staggeringly since late 2025. For instance, Anthropic (Claude) is up about 175%, OpenAI (ChatGPT) is up over 70%, Alphabet (Gemini) is up about 26%, and Google (DeepMind) is up 61%. Most of these have a knack for generating debt without generating a penny of profit.
It's worth being precise here: this is not a market-wide consensus that AI is a dead end. Serious analysts argue the pullback is a healthy repricing after an extraordinary run-up, not a verdict on the technology's usefulness. They assert that, unlike the dot-com era, the largest AI spenders are funding buildouts from operating cash flow rather than debt.
But the fact that the debate is this unsettled, this far into the adoption cycle, is informative in itself. Two years ago the assumption embedded in AI valuations was that reliability and return would simply follow scale. That assumption is now being tested publicly, in real time, against actual earnings reports.
Public Is Rejecting the Infrastructure, Not Just the Valuations
Perhaps the clearest evidence that AI's social license is eroding – separate from anything happening in a spreadsheet or a stock ticker – is what's happening at the local zoning level. Data Center Watch recorded at least 75 U.S. data center projects, worth roughly $130 billion, blocked or delayed by local opposition in the first quarter of 2026 alone, which matches the entire prior year's total in a single quarter.
Separate tracking from Heatmap News counts at least 20 outright project cancellations in that same window, representing $41.7 billion in stalled investment and 3.5 gigawatts of demand, with organized opposition groups more than doubling to over 800 across 49 states.
These aren't fringe fights. A major Virginia project near Manassas National Battlefield Park collapsed after a court invalidated the county's zoning approvals; New York enacted a state-level moratorium on new large-scale data center permits; Tulsa's city council froze new construction for nine months after residents raised concerns about siting in underserved neighborhoods.
A Brookings analysis of the trend makes a really good observation worth sitting with: data centers have become the most visible, tangible manifestation of AI that ordinary citizens can actually see and organize against, and opposition to them functions as a proxy for a much broader, harder-to-litigate anxiety about AI's effect on jobs, energy costs, and daily life. Americans may not know how to resist AI in the abstract, but they certainly do know to shut down efforts to build infrastructure in their backyard. Gallup found roughly 70% of Americans oppose having a data center built in their area, and a separate Heatmap Pro poll put opposition at 75%.
Data Center Opposition
I think the opposition should be seen as a genuinely contested empirical picture, not a one-sided one. Some energy researchers argue that the link between a specific data center and a specific household's electric bill is harder to establish than opponents claim, and that certain objections have become rallying points before the underlying evidence has been verified. But whether or not each claim holds up, the pattern of opposition itself is the significant fact.
Public opposition has many causes. I think the following are among the most contentious.
- Electricity Costs and Grid Strain
Data centers are large, constant electricity draws that can force utilities to build new generation, substations, and transmission lines. The reality is that these costs may land on ordinary ratepayers' bills unless regulators carve out special rate structures. U.S. data centers already consume roughly 4% of national electricity, a share projected to more than double by 2030, and I have read that in at least one regional capacity market the added demand was linked to a $9.3 billion increase in costs.
- Water Consumption
Cooling systems for large server farms can consume millions of gallons of water annually, a particular flashpoint in drought-prone regions. States, including California, have begun requiring data centers to disclose and certify their water use as a condition of local licensing.
- Noise and Light Pollution
Continuous mechanical noise from cooling equipment and backup generators, as well as overnight lighting, are among the most frequently cited quality-of-life objections at planning and zoning hearings.
- Property Values and Land Use
Residents near proposed sites regularly cite fears of declining home values, loss of green space and farmland, and a change in community character due to large, industrial-scale structures.
- Air Pollution
In areas where utilities respond to new demand by running natural gas "peaker" plants more often, or where data centers install on-site diesel backup generators, opponents point to localized air pollution as a health concern.
- Equity and Site Fairness
Several fights, such as the one in Tulsa, have centered on data centers being sited disproportionately in lower-income or historically underserved neighborhoods, echoing older environmental-justice objections to industrial development. I have seen a protest sign suggesting that data centers should be built next to the state's capital.
- Thin Local Benefit
Data centers generate meaningful tax revenue but comparatively few permanent jobs once construction ends, leaving some communities questioning whether the trade-off – decades of infrastructure investment for a facility that employs a few dozen people – is worth it.
- Process Opacity
Developers have frequently sought non-disclosure agreements during site negotiations and used shell-company names to acquire land. This practice has itself become a source of local distrust and is now the target of proposed state legislation that would ban NDAs in data center permitting.
___________
This article, The Day of AI Reckoning: Part II, published on September 16, 2026, is authored by Jonathan Foxx, PhD, MBA, the Chairman & Managing Director of Lenders Compliance Group, founded in 2006, the first and only full-service mortgage risk management firm in the United States, specializing exclusively in residential mortgage compliance.
