The Yellow Line: How a 1939 Mortgage Grade Still Shapes West Orange, NJ

The Yellow Line: How a 1939 Mortgage Grade Still Shapes West Orange, NJ
West Orange, New Jersey

The Yellow Line

I am perhaps one of a unique group of people who made a home purchase in a data-informed way. From proximity to highways, flood risk, neighborhood stability, fatal police encounters, and walk scores, and don't even get me started on radon (perhaps in another post). Given the structural inequities in the United States, I was cognizant that the effects of policies long since discontinued still reverberate in hidden ways, shaping individual and community wellbeing. This data storytelling post explores how a 1939 mortgage grade still determines who lives well in West Orange, and why the neighborhood low on the radar in redlining stories might be the one that matters most.

$40,223 – $250,000+ The range of median household income across block groups inside a single West Orange zip code, 07052.[1]
Scroll
What the grades meant

Redlining stories usually start with red. This one starts with yellow.

In the late 1930s, the federal Home Owners' Loan Corporation sent appraisers into neighborhoods across the country and graded them for mortgage risk. Four grades, color-coded: A, green, "best." B, blue, "still desirable." C, yellow, "definitely declining." D, red, "hazardous."

The grade was supposed to reflect housing quality. In practice it weighed just as heavily on who lived there: race, ethnicity, income, and how long the appraiser expected that to stay the same.

Most retellings of this history go straight to the red D neighborhoods. The yellow C grade, "definitely declining" and one notch above hazardous, gets skipped past, even though it covered more ground and, as the research later in this piece shows, may have done more lasting damage.

1939 HOLC residential security map of Essex County, New Jersey, showing neighborhoods color-coded green, blue, yellow, and red
Essex County, New Jersey, 1939. Source: Home Owners' Loan Corporation, digitized by the Digital Scholarship Lab, "Mapping Inequality: Redlining in New Deal America," University of Richmond.
A: Best, 20% of the county B: Still desirable, 50% C: Definitely declining, 21% D: Hazardous, 9%
Zooming into West Orange

Area C-12: Pleasantdale

West Orange's Pleasantdale neighborhood was graded C-12, yellow, not red. Here is what the appraiser wrote about it on January 1, 1939.

Grade C, Yellow

C-12: West Orange, Pleasantdale

"Level to rolling. Area lies in valley to west of Orange Mountains. Topography is attractive. This development, while fairly new, was of somewhat poor calibre. Facilities of all kinds are limited... The section is about 30% built up. However, at the present time, it does not deserve better than a third grade plus."
Negro population: 0% Population: majority Jewish Class: mixed, white collar Est. income: $1,000–2,000

Source: HOLC Area Description C-12, Mapping Inequality: Redlining in New Deal America, Digital Scholarship Lab, University of Richmond.

Not who you'd expect

C-12 wasn't downgraded for the reason you'd assume

In 1939, the percentage of Black Pleasantdale residents was 0%. As of the 2020 Census, the same neighborhood is about 22.9% Black, alongside 53.2% White and 11.8% Hispanic or Latine[2], a very different picture from what the appraiser recorded. The appraiser's 1939 complaint was about construction quality and, unmistakably, about the neighborhood being Jewish. Across town, a different Essex County neighborhood was given the county's lowest grade, for a different reason entirely.

Grade C, Yellow

C-12: West Orange, Pleasantdale

"...of somewhat poor calibre. Facilities of all kinds are limited... does not deserve better than a third grade plus."
Negro: 0%Population: majority Jewish
Grade D, Red

D4: Newark, Third Ward

"This is Newark's worst slum section... About 45% of the structures are tenements and low-grade flats. Considerable demolition and boarding-up has taken place... useful only to those in lowest income brackets."
Negro: 70%Shifting/infiltration: many on relief

Same appraisal system, same year, same county. One neighborhood was marked down a notch for ethnicity and housing stock. The other was written off entirely, on the basis of race.

The research

The yellow line, not the red line, did the most damage

Economists at the Federal Reserve Bank of Chicago[3] tested this by comparing neighborhoods that sit on opposite sides of a HOLC grade boundary: a block graded B against the block next to it graded C, or a block graded C against the block next to it graded D.

The C to B boundary, the yellow line, showed larger and more persistent effects on homeownership, home values, and segregation than the D–C boundary, the actual red line. Their explanation: the yellow grade was the first official signal that a neighborhood was declining. It triggered tighter credit and White flight earlier. By the time a neighborhood reached D, it had often already been written off. The marginal damage of the red label was smaller because, as the authors posit, there was less left to lose.

B: Blue

"Still desirable"

C: Yellow

"Definitely declining," the first official signal of decline

The C–B boundary: larger, more persistent effects
C: Yellow

Already discounted by lenders

D: Red

"Hazardous," but often already disinvested

The D–C boundary: smaller marginal effect

Across the country, HOLC maps account for an estimated 15–30% of the enduring gap in neighborhood demographics (read: Black population percentage) and Black homeownership rates between 1950 and 2010, a substantial share of it driven by yellow-line boundaries, not red ones.[3]

Same county, different fates

Four neighborhoods, one 1940 map

Essex County's grades tracked race and class from the very top of the scale to the bottom.

A: Glen Ridge, Ridgewood Ave. North

"A first grade suburban area which is still growing"

0% Negro. White collar business executives. Est. income $5,000–50,000.

B: Glen Ridge, North Side

"Formerly a desirable section in a very high grade residential town"

0% Negro. White collar. Est. income $2,500–10,000.

C: Bloomfield

"Slowly going down-hill"

25% foreign families, Italian. "The few negroes are concentrated on John St." Est. income $750–2,000.

D: Newark, Third Ward

"Newark's worst slum section"

70% Negro. Est. income $500–1,500. Slum clearance underway.

Try it yourself

Pick a want. Live the life attached to it.

For illustrative purposes, below are three composite couples; however, everything about their starting point is real. Each partner's race, ethnicity, occupation, and income is derived directly from the actual 1939 area descriptions for A22, C-12, and D4 quoted earlier in this piece. Pick one of the three wants below, nothing else, and you'll walk that couple's life forward from 1940. Who they are gets revealed only after you've made your choice, on purpose: in 1940, nobody chose their grade either.

The income gap between the Sinclairs and the other two families did not start with this map. Executive suites and most licensed professions ran an explicit or de facto color bar and, for Jews, informal quotas, so the same ambition and ability sorted people into very different pay by the time they reached 1939. Many unions barred Black workers outright or paid them a separate, lower wage for the same job. The New Deal labor protections passed just a few years earlier, the ones that set minimum wages and union rights, explicitly excluded domestic and agricultural work, the jobs where most Black workers in America, and most working Black women in particular, were concentrated. Historians who have studied the legislative record, most notably Ira Katznelson, describe this as a documented political bargain, not an oversight: Southern Democrats in Congress made the exclusion a condition of their support for the broader New Deal, since those two job categories covered roughly 60% of the nation's Black workforce at the time (some scholars attribute part of the exclusion to tax-administration practicalities instead, so this is the dominant reading, not an uncontested one). And a separate barrier ran under all three families regardless of grade: married women in the U.S. generally could not hold a mortgage or line of credit in their own name until the Equal Credit Opportunity Act of 1974, a barrier so normalized at the time that later research on the women who fought to change it found many hadn't experienced it as discrimination so much as simply how things were. The mortgage system did not create any of these gaps. It compounded ones that already existed.

W+E

William & Eleanor Sinclair

Aged 36 and 33. Glen Ridge, Ridgewood Ave. North. Area A22, Grade A, Green.

Race: White. William is a business executive in Newark. Eleanor ran the books for a college alumnae fund before she married, now manages the household and two kids. Household income around $12,000 a year; the home itself, the mortgage, and the deed are all in William's name; New Jersey law let married women own property by 1940, but mortgage lending was a separate matter, decided by the banks themselves, not the courts.

Want: William, to keep growing his firm and buy a second property. Eleanor, to go back to paid work once the kids are older, though she can't yet picture a clear way back.

Worry: William, that the town's zoning might not hold against "the wrong element" moving in from the west. Eleanor, a harder thing to put a name to: that keeping a fine house and raising accomplished children is supposed to be the whole of a full life, and that most days she believes it.

R+D

Ruth & David Adler

Aged 29 and 32. Pleasantdale, West Orange. Area C-12, Grade C, Yellow.

Race: White; religion: Jewish. David is a bookkeeper. Ruth sews for neighbors and dreams of a shop with her own name on the door, though any mortgage will carry David's name alone. One young son. Household income around $1,500 a year.

Want: a home of their own, out of a crowded Newark walk-up. Ruth's dress shop, David's steady advancement at the firm.

Worry: David, that being Jewish will always cost something, even with steady work and good credit. Ruth, that if the shop ever does open, the loan behind it, and probably the credit for the idea, will carry David's name, same as everything else does.

J+V

James & Vivian Coleman

Aged 31 and 28. Third Ward, Newark. Area D4, Grade D, Red.

Race: Black. James works as a longshoreman on the Newark docks, hired by the day since the union local won't let Black men into steady gang work. Vivian cleans houses in Glen Ridge and Montclair, the kind of work the New Deal's new labor protections didn't cover at all. Two kids. Household income around $1,000 a year, combined.

Want: James, to buy instead of rent and eventually open a repair shop. Vivian, steady enough hours to be home when the kids are.

Worry: James, that no bank will ever say yes, no matter how much he saves. Vivian, that the domestic work she does was left out of the New Deal's new protections on purpose, and that everyone she knows is well aware of it, even if nobody outside their neighborhood seems to be.

Beyond property values

The grade shows up in bodies, not just deeds

National meta-analyses of redlined (D-grade) neighborhoods find 41% higher odds of preterm birth, elevated cardiovascular disease, asthma, and firearm injury, and roughly a 4.8-year gap in life expectancy between the best- and worst-graded tracts.[4]

The yellow grade shows up too, independently. A Detroit-area study found significantly worse neighborhood determinants-of-health scores in yellow-lined tracts compared to blue- and green-lined tracts, not just red compared to everything else.[5] A Massachusetts study found that residents of yellow-lined areas independently faced an elevated risk of late-stage breast and lung cancer diagnoses, a risk layered on top of, and separate from, the elevated risk already documented for red-lined areas nearby, holding even for residents of neighborhoods that are affluent today.[6]

Grade A/B
baseline
Grade C
worse
Grade D
worst

Illustrative, not a specific dataset. It summarizes a consistent direction across the Detroit Determinants of Health Index study[5] and the Massachusetts late-stage cancer study.[6]


Did the grade shape where the highway went?

In Essex County, the answer splits in two, depending on where you look.

Newark: Third Ward (D4)

Grade and route point the same way

I-280 was cut directly through Newark's D-graded Central and North Wards, displacing at least 5,000 residents alongside I-78's construction. Planning language from the era described the goal plainly: highways to carry suburban commuters "quickly and safely over the slums."

West Orange: Pleasantdale (C-12)

Terrain, not the grade, drove the route

I-280 runs directly along Pleasantdale's southern border. Exit 7 sits at the neighborhood's edge, but the West Orange segment was an engineering fight with the Watchung Mountains, moving roughly 4 million cubic yards of rock. Pleasantdale itself was never cleared for it. It's intact today, bordering the highway, and it's one of the wealthiest neighborhoods in the region.

Both are true, and they shouldn't be collapsed into one claim. A few miles apart, the same interstate did two very different things to two differently graded neighborhoods.


The grade shows up in flood maps and police data too

Two more data points a house hunter might check today, since the grade reaches into both. Comparing nearly identical properties on opposite sides of the same HOLC boundary, researchers at the Federal Reserve Bank of Richmond found flood risk rises in conjunction as the grade worsens from A to D, evidence that lower-graded neighborhoods received less of the drainage and green-infrastructure investment that limits flood exposure, not that they were simply built on worse ground.[7] And a national study of fatal police encounters between 2000 and 2020 found tracts that were redlined saw 66% more fatal encounters than tracts graded A, a gap that held even after accounting for today's demographics and poverty rate.[8]

Sources: Phan, Conzelmann, Salazar-Miranda, and Hoffman, Federal Reserve Bank of Richmond, "Long-Term Effects of Redlining on Environmental Risk Exposure," 2022, on flood risk. Mitchell and Chihaya, Social Science & Medicine, "Tract Level Associations Between Historical Residential Redlining and Contemporary Fatal Encounters With Police," 2022, on policing.

Who lived in the yellow zones

Nationally, the yellow and red zones were disproportionately Black and Latine

A 2024 study linking HOLC maps to the nationally representative Add Health cohort found Black and Latine adolescents were several times more likely than White adolescents to have grown up in a formerly yellow- or red-lined area, and those who did went on to attain lower socioeconomic status as adults.[9]

Latine
28%
Black
22%
White
8%

Share of adolescents who grew up in a formerly yellow- or red-lined urban area, by race/ethnicity. Source: Add Health cohort study, 2024.

Pleasantdale doesn't fit that pattern. Which raises the question this piece can't fully resolve.

West Orange today

The old C-12 is now one of the wealthiest places in America

Pleasantdale today ranks among the top 15% of neighborhoods nationally by income, higher than 87.9% of neighborhoods in the United States, and just 0.8% of children there live below the federal poverty line, a lower rate than in 79.1% of American neighborhoods, according to NeighborhoodScout's proprietary neighborhood estimates.[10]

$40,223West Orange, 07052$250,000+

That outcome runs counter to what the national yellow-line research would predict. Pleasantdale isn't necessarily an exception to that research, though; it may be a later chapter of the same story. Land that was disinvested for decades because of its yellow grade can still become desirable again, and when it does, whoever can afford to move in captures the gain, the same dynamic now driving gentrification and displacement pressure in other formerly yellow-lined neighborhoods around the country.

One caveat, not a competing theory: the "decline" the 1939 appraiser predicted for Pleasantdale because it was a Jewish neighborhood never fully materialized, unlike the outcome for the redlined tracts nearby. Both things are true at once. This piece doesn't try to resolve the tension between them, only to name it.

The one number both cards recorded

How close does each place still match its 1939 rating?

The HOLC form captured exactly one racial statistic, and it recorded the same value for both areas: "Negro: 0%." That's not an estimate or a proxy. It's the literal documented baseline both places were graded on. Comparing each against that same number, eighty-one years later, shows how far each has actually moved from what it was rated for being.

West Orange · Pleasantdale, Area C-12 · Grade C, 1939
1939, documented0% Negro
2020 Census, Pleasantdale CDP22.9% Black
Drifted ~23 points from the exact baseline it was graded on.
Glen Ridge · Areas A22 & B36 · Grade A/B, 1939
1939, documented0% Negro
2020 Census, Glen Ridge borough3.7% Black
Drifted ~4 points, still close to what it was rated for being.

Sources: 2020 Decennial Census, Demographic Profile (DP1), Pleasantdale CDP and Glen Ridge borough, Essex County, NJ. Pleasantdale CDP is the closest present-day geography to the 1939 C-12 boundary; Glen Ridge borough is coterminous with the town graded in 1939, so no comparable boundary mismatch applies there.

Two grades, two towns, today

Pleasantdale (Grade C) and Glen Ridge (Grade A/B), side by side

West Orange was marked down a notch in 1939. Glen Ridge, a few miles north, was graded the county's best, in part explicitly because it was "highly restricted" and reported 0% Black residents. Eighty-five years on, one neighborhood grew far more mixed than what its grade predicted; the other has stayed close to what its grade rewarded it for being.

Pleasantdale · Grade C, 1939

White
53.2%
Black
22.9%
Hispanic
11.8%
Asian
7.5%
Multiracial
3.7%
Other
0.9%

Glen Ridge · Grade A/B, 1939

White
74.7%
Black
3.7%
Hispanic
7.2%
Asian
7.5%
Multiracial
6.3%
Other
0.6%

Demographic figures: 2020 Decennial Census, Pleasantdale CDP and Glen Ridge borough, matched geographies, same vintage, same six categories on both sides (all figures sum to 100%). Income: Pleasantdale (Area C-12) is $129,753, but that figure is West Orange township-wide, not Pleasantdale-specific (Pleasantdale is one neighborhood inside a much larger township), so treat it as directional. Glen Ridge's $250,000+ is borough-wide and borough-specific, since Glen Ridge is small enough that the whole municipality maps closely to its 1939 area.

Not an isolated case

West Orange wasn't the only one

Marking a neighborhood down because Jewish families lived there was a documented, repeated appraisal practice, not a West Orange quirk.

Boston: Mattapan, Roxbury, Dorchester

Graded C9, C10, C11

Marked yellow explicitly over Jewish "infiltration."

South Philadelphia

HOLC appraiser notes

"Infiltration of Jewish into area have depressed values."

Los Angeles: Pico-Robertson

Downgraded to "low yellow"

Cited "definite infiltration of Jewish families."

West Orange: Pleasantdale

C-12, "third grade plus"

0% Negro, majority Jewish, "somewhat poor calibre."

Zooming back out

The gap holds whichever neighbor you pick

Two ways to see how far the gap reaches, using two different towns near Glen Ridge: one graded in this same 1939 map and built up throughout this piece, the other not HOLC-graded at all but literally next door.

A few miles apart, both graded in 1939

Glen Ridge and West Orange sit a few miles apart in the same county, graded A/B and C respectively. On the same 2020–2024 ACS vintage used throughout this piece, Glen Ridge's median household income is $250,000 or more, the top-coded ceiling of what the Census Bureau reports. West Orange township's is $129,753, roughly half.[11]

Glen Ridge
$250,000+
West Orange
$129,753

Both figures are municipality-wide, 2020–2024 ACS 5-Year Estimates, same geography level and vintage. (This is the township-wide West Orange figure, not the Pleasantdale-specific numbers used in the 1939-baseline comparison earlier in this piece.)


Actually next door: Glen Ridge and East Orange

East Orange doesn't appear elsewhere in this piece (it wasn't one of the four Essex County areas quoted from the 1939 map earlier), but it's worth naming because it's not a few miles from Glen Ridge, it's the town Glen Ridge directly borders. Glen Ridge's median household income is the same $250,000 or more. East Orange's is $60,830.[13] The line between them isn't a metaphor; it's a literal municipal boundary.

Glen Ridge
$250,000+
East Orange
$60,830

Both figures municipality-wide, 2020–2024 ACS 5-Year Estimates. East Orange was not one of the four HOLC-graded Essex County areas quoted earlier in this piece, so this comparison stands on its own as a present-day border effect, not a 1939-grade comparison.

Countywide, Black and Latine households in Essex County are half as likely to own a home as White households, and three times more likely to live in poverty.

The grade never disappeared.
It just stopped being drawn on official maps.

And the label that seems to have done the most lasting damage wasn't even the red one.

What this piece does not answer yet

A more precise version of this story is possible

Everything here uses public, archival, and published sources. A sharper follow-up would merge the actual C-12 and neighboring HOLC boundaries with current census-tract data for West Orange, East Orange, and Orange, then run the same boundary-design comparison the Chicago Fed used, this time on homeownership, home values, and health outcomes specific to this county rather than the national picture.

That analysis does not exist here. It is the natural next project, not a promise, just a note on where the evidence trail would go next.

How this was built

This piece draws on primary HOLC area-description documents (Mapping Inequality: Redlining in New Deal America, Digital Scholarship Lab, University of Richmond), peer-reviewed meta-analyses and systematic reviews on redlining and health, Federal Reserve Bank of Chicago working papers on HOLC boundary effects, a nationally representative adolescent cohort study, and public income, poverty, and health-indicator data for West Orange and Essex County.

Where the piece interprets rather than reports, most notably the two readings of Pleasantdale's present-day wealth in the section above, that interpretation is flagged explicitly. It is not settled research. Every other claim traces to a specific document, dataset, or peer-reviewed study, listed below.

Sources

  • Home Owners' Loan Corporation area descriptions, Essex County, NJ (1939-40): Mapping Inequality: Redlining in New Deal America, Digital Scholarship Lab, University of Richmond
  • Aaronson, D., Hartley, D., & Mazumder, B. (2017, revised 2020). The effects of the 1930s HOLC "redlining" maps (Working Paper No. WP-2017-12). Federal Reserve Bank of Chicago
  • Federal Reserve Bank of Minneapolis, conference materials on HOLC boundary-design research
  • Harris, K. M. (2024). Traces of historical redlining in the contemporary United States: New evidence from the Add Health cohort. Population Research and Policy Review. University of North Carolina at Chapel Hill
  • Mehdipanah, R., McVay, K., & Schulz, A. (2023). Historic redlining practices and contemporary determinants of health in the Detroit metropolitan area. American Journal of Public Health
  • Krieger, N., et al. (2020). Cancer stage at diagnosis, historical redlining, and current neighborhood characteristics: Breast, cervical, lung, and colorectal cancers, Massachusetts, 2001–2015. American Journal of Epidemiology
  • Lee, Donley, Gill et al., systematic review and meta-analysis, redlining and health outcomes, Social Science & Medicine (2022)
  • Case Western Reserve University systematic review of 36 studies, historic redlining and health across the lifespan (2024)
  • Newark Changing (Myles Zhang), "Redlining & Manufactured Decline," I-280/I-78 displacement figures and Newark Third Ward (D4) routing
  • AARoads / Interstate-Guide, Interstate 280 (New Jersey) construction history: Watchung Mountains engineering and Pleasantdale/Exit 7 location
  • Phan, T., Conzelmann, C., Salazar-Miranda, A., & Hoffman, J., "Long-Term Effects of Redlining on Environmental Risk Exposure," Federal Reserve Bank of Richmond, Working Paper 22-09 and Economic Brief No. 22-51 (2022), on HOLC grade and present-day flood and heat risk
  • Mitchell, J., & Chihaya, G. K., "Tract Level Associations Between Historical Residential Redlining and Contemporary Fatal Encounters With Police," Social Science & Medicine (2022)
  • Davis, C. J., "Shape or Fight? New York's Black Longshoremen, 1945 to 1961," on discrimination against Black dockworkers in the New York/Newark harbor
  • Katznelson, I., Bateman, D., & Lapinski, J., on the exclusion of agricultural and domestic workers from Social Security Act (1935) protections as a negotiated concession to Southern Democrats; Social Security Administration's own historical account in the Social Security Bulletin offers a contrasting administrative-feasibility reading
  • Thurston, C., At the Boundaries of Homeownership, and Marketplace/APM, "50 years ago, it was legal to deny a woman credit without a male co-signer" (2024), on NOW's pre-ECOA testimonial archive, including a documented New Jersey account
  • "Married Women's Property Acts in the United States," on New Jersey's 1852 and 1874 statutes establishing married women's independent property rights
  • Rutgers New Jersey State Policy Lab, "The Impact of City Borders on Achieving Health Equity" (2024)
  • The Jersey Bee, "In Essex County, New Jersey's History of Segregation Persists" (2024), citing New Jersey Institute for Social Justice
  • Vital City NYC, "The Stains of Redlining," Jacob William Faber, New York University (2023)
  • Institute for Housing Studies, DePaul University, "Following the Yellowlined Road"
  • NeighborhoodScout, Pleasantdale, West Orange, NJ neighborhood profile, retrieved August 7, 2026: income percentile and child poverty rate (ages 17 and under). Proprietary neighborhood estimate, not a directly reported Census table; percentile rankings may shift as the provider's model updates.
  • World Population Review, Pleasantdale, New Jersey, retrieved August 7, 2026: overall poverty rate estimate. Proprietary estimate, not a directly reported Census table.
  • U.S. Census Bureau, QuickFacts: Newark city, New Jersey, Median Household Income and Persons in Poverty, percent, 2020–2024 American Community Survey 5-Year Estimates
  • Advocates for Children of New Jersey, Newark Kids Count 2024: A City Profile of Child Well-Being (contextual source on Newark child and family well-being indicators)
  • U.S. Census Bureau, QuickFacts: Glen Ridge borough, New Jersey, Median Household Income, 2020–2024 American Community Survey 5-Year Estimates
  • U.S. Census Bureau, QuickFacts: East Orange city, New Jersey, Median Household Income, 2020–2024 American Community Survey 5-Year Estimates
  • U.S. Census Bureau, QuickFacts / Census Reporter profile, West Orange township, Essex County, New Jersey, Median Household Income, 2020–2024 American Community Survey 5-Year Estimates
  • U.S. Census Bureau, American Community Survey 2019–2023 5-Year Estimates, Table B19013 (Median Household Income), block group level, West Orange township, Essex County, NJ (Census Tracts 173.01, 173.02, 174, 175, 176, 177, 178, 179, and 180, comprising ZIP 07052), accessed via api.census.gov
  • U.S. Census Bureau, Historical Census of Housing Tables: Home Values, 1940–1960 median home value by decade
  • "Pleasantdale, New Jersey," Wikipedia, citing U.S. Census Bureau, 2020 Decennial Census Redistricting Data (Public Law 94-171) and Demographic Profile (DP1) for Pleasantdale CDP
Next
Next

The Hidden Cost of Progress: Why Your Utility Bill Is Subsidizing the AI Revolution