The Backlash Clock: A Narrative

Two booms, one backlash, ninety-nine years apart. JD Unfiltered Research. Analog reasoning, not investment advice. Historical parallels are illustrative frameworks, not forecasts.

I. The setup: nobody revolted against the machine

In neither the 1920s nor the 2020s did the public revolt against the technology itself. Adoption was voracious both times. When the Commerce Department licensed KDKA in October 1920 and its Harding–Cox election broadcast went out that November, radio became a craze that swept the country and turned into a vast new industry. Ninety-nine years minus a few weeks later, ChatGPT launched on November 30, 2022 and reached 100 million monthly users in two months, the fastest-growing consumer application in history.

The backlash, in both eras, targeted something else entirely: who owned the technology, what it cost, how it was financed, and what it did to culture.

In the 1920s that meant three overlapping fights. The cultural one played out in venues like The Forum's 1929 debate, where novelist Jack Woodford predicted advertising would leave radio "as dead as a Democrat" within two years and warned of passive, anesthetized listeners, while RCA's Gen. James Harbord answered that the radio listener was "free from the contagion of the crowd." The money fight centered on Samuel Insull's $4 billion, 32-state pyramid of holding companies serving some 15 million people, which collapsed in 1932 and wiped out 600,000 shareholders. And the propaganda fight targeted the National Electric Light Association's campaign to subsidize teachers, ghostwrite editorials, and place utility-authored material in school textbooks.

The 2020s rhyme on every count. Doomscrolling and algorithmic-feed anxiety replace the "dazing, almost anesthetic effect upon the mind" that Anne O'Hare McCormick attributed to radio. Justice Department antitrust probes of the AI compute stack replace the power-trust inquiries. And public concern about AI has climbed from 37 percent in 2021 to 52 percent in 2026, with job-loss fears rising fastest: 79 percent of Americans now tell Gallup that AI will reduce jobs within a decade.

II. The question: did 1927 see a "midterm threat" like 2026's?

On the 99-year clock, 2026 sits opposite 1927. The question is whether the political storm now bearing down on the November 2026 midterms, the one about AI, data centers, and electricity bills, had a genuine counterpart in 1927.

The short answer is yes, and the match is unusually tight. 1927 is precisely the year the power trust stopped being a business-press story and became an organized political project. But there is one honest wrinkle in the analog, covered in Section V: the electoral payoff arrived on a slight delay.

III. 2026: the backlash reaches the ballot

By August 2026, AI is not a niche issue. It is on the ballot. A Washington Post analysis of roughly 1,200 candidate websites found AI or data-center policy statements in nearly 40 percent of races nationwide, more than mentions of Israel, manufacturing, or racism. Gallup finds about 70 percent of Americans, including 63 percent of Republicans, oppose new hyperscale data centers in their area. The share of Americans saying data centers raise their electricity bills jumped from 43 percent in January to nearly 60 percent in August.

The early returns already drew blood. In November 2025, Virginia's Abigail Spanberger won the governorship pledging to make data centers pay their fair share of grid costs, and Georgia Democrats ousted two Republican incumbents from the Public Service Commission, the party's first wins for that body since 2007, after six rate hikes in two years pushed typical bills to about $175 a month. Winning candidate Peter Hubbard's message was pure ratepayer populism: data centers "are being offered a sweetheart deal of five cents a kilowatt-hour, while you and I, residential customers, pay 21 cents."

Now the wave points at November 2026. In Ohio, Sherrod Brown has made data centers the centerpiece of his Senate campaign, branding incumbent Jon Husted "the face of data centers in Ohio"; an internal NRSC memo calls the issue a "sleeper issue for the entire election cycle" and warns that data centers are the anchor hanging around Husted's neck. In Michigan, Senate nominee Abdul El-Sayed built a primary win around an "AI Under Democracy" platform calling for public ownership of AI companies, an automation tax, and an AI dividend.

Even Republicans are breaking ranks. Texas Gov. Greg Abbott suspended data-center projects pending an audit, drawing Trump's public rebuke that the move was a mistake, and Pennsylvania's Josh Shapiro signed an August 18 executive order forcing developers to carry their own generation and transmission costs. Overhead, the federal government has picked its side: Executive Order 14365, signed December 11, 2025, created a Justice Department task force to sue states over AI laws that conflict with a national deregulatory framework, and the order is itself now a campaign flashpoint.

Rep. Greg Casar's August warning distills the moment. This backlash, he said, "is just the beginning of the political earthquake that is going to come."

Now run the clock back ninety-nine years.

IV. 1927: the year the power trust entered politics

February 23, 1927. Coolidge signs the Radio Act, born of monopoly fear. Its Section 13 ordered the new Federal Radio Commission to strip the license of any company convicted of "unlawfully monopolizing or attempting to monopolize radio communications," an anti-air-trust clause aimed largely at RCA. Co-author Sen. Clarence Dill named the anxiety plainly: "The great feeling about radio in this country is that it will be monopolized by the few wealthy interests."

March 11, 1927. Sen. George Norris convenes power-trust opponents, including Gifford Pinchot, in his office to plan a Senate investigation of the electric industry.

June and December 1927. Sen. Thomas Walsh, fresh from prosecuting Teapot Dome, introduces and reintroduces resolutions to investigate electric utilities. By December the proposal is winning newspaper support nationwide, and the industry pivots from opposing the probe to steering it toward the friendlier FTC.

December 5 and 6, 1927. The Boulder Dam bills land in Congress, forcing the public-versus-private power question onto the national agenda after years of seven-state wrangling.

The dam bursts weeks past the analog year. On February 15, 1928, after a nine-hour fight, the Senate votes 46 to 31 to order the FTC investigation. Senate Resolution 83 directed the FTC to examine not just holding-company finances but whether utilities had spent money "to influence or control public opinion" and, since 1923, to influence or control elections. The hearings ran from March 1928 to December 1935 and produced 63,000 pages of testimony across 278 utility companies. Walsh took the findings straight to the electorate, charging that utility propaganda had been "introduced in the schools for the purpose of poisoning the minds of our youth."

By the 1928 presidential race, power was a top-line issue. Al Smith ran on government development and ownership of hydroelectric sites. Hoover branded it state socialism. And Norris, a Republican, crossed party lines to endorse Smith over public power, prefiguring his 1932 support for FDR, the Portland "Insull monstrosity" speech, and ultimately the TVA.

V. The verdict: same storm, slightly different clock

So did 1927 see the same kind of rising political threat? Structurally, yes, almost beat for beat.

An infrastructure "trust" dragged into accountability hearings. Walsh's 1927 resolutions and S. Res. 83 mirror 2026's demands that AI firms disclose costs, safety practices, and lobbying, down to the shared suspicion that the industry is buying elections. S. Res. 83 literally ordered an investigation of utility money in elections. In 2026, AI-aligned PACs had put $185 million into the midterms by March.

Ratepayer anger as the trigger. Hubbard's sweetheart-deal line in Georgia is the 1920s grievance verbatim: households subsidizing the trust's expansion while paying monopoly rates.

Insurgents crossing party lines against an industry-friendly executive. Norris endorsing Smith over Coolidge-Hoover orthodoxy parallels Abbott and Byron Donalds breaking from Trump's pro-data-center stance in 2026.

The executive shielding the industry. Coolidge's 1928 pocket veto and Hoover's 1931 veto of Muscle Shoals, in which Hoover wrote that he was "firmly opposed to the Government entering into any business" in competition with citizens, rhyme with EO 14365's preemption of state AI laws.

Whoever controls it controls the nation. Edward Nockels's 1928 warning about radio, that whoever controls radio broadcasting in the future will eventually control the nation, is El-Sayed's public-control argument a century early.

The honest caveat: the 1926 midterms themselves were not fought over the power trust. They turned on farm distress and the McNary-Haugen veto. In the 1920s, 1927 was the year of political mobilization, the meetings and resolutions and bills, while the ballot-box consequences arrived in 1928 in the presidential race, in 1930 through 1932 with Insull's collapse and FDR, and in 1935 with PUHCA. In the 2020s, the mobilization and the electoral punishment are happening simultaneously, in 2026 itself, with early casualties already recorded in Virginia and Georgia in 2025.

That compression matches a pattern already flagged on the dashboard's timeline: the modern regulatory wave is running roughly a decade ahead of its 1920s analog. Survey data, cable news, and primary challenges transmit grievance faster than the Congressional Record ever did. If the analog holds, 2026 is 1927 with the electoral fuse shortened, the year the anti-trust coalition organizes, finds its Norrises and Walshes, and discovers the issue wins races. The 1920s version of that discovery ended, seven years later, in a death-sentence clause for the holding companies. The open question for the 2020s is whether the shortened fuse also shortens the distance to the modern equivalent.

VI. Did the backlash hit the tape? In 1927, not until it hit all at once

If politics was organizing against the trusts in 1927, the stock market never got the memo. The Dow returned 28.75 percent in 1927 and 48.22 percent in 1928 while Walsh's resolutions and the Radio Act's monopoly clause were live. The politically targeted sectors were the very ones leading the boom. Utilities gained roughly 855 percent from 1924 to their September 1929 peak and made up about 18 percent of NYSE value at the top, while RCA, the named target of the Radio Act's anti-monopoly Section 13, ran from about $85 in early 1928 to a $549 peak in September 1929. When the Senate voted 46 to 31 on February 15, 1928 to send the power probe to the FTC, the industry read it as a victory. Power interests congratulated themselves, TIME reported.

The political risk stayed unpriced until a regulator with rate authority acted. Harold Bierman's research identifies the Massachusetts Department of Public Utilities' October 11, 1929 refusal to let Boston Edison split its stock, declaring the shares not worth their price, as the crash's proximate trigger. The stock fell from a $440 high toward $299, and the leveraged utility pyramids unraveled from there. Utilities then fell hardest of any sector, down 55 percent against 48 percent for industrials in the crash months, and the Dow Utility Average ultimately lost 92.67 percent, not bottoming until 1942. Insull's securities went from a $500 million gain in the 50 days before August 23, 1929 to receivership by April 1932. Politics finished the job later: event studies show holding-company values moving with each PUHCA legislative milestone in 1935.

2026 is running the same story with one structural difference. The market is pricing the politics in real time. Abbott's Texas grid-audit order on August 4 knocked Vistra down 7 percent and NRG as much as 16 percent the same day. Shapiro's Pennsylvania order on August 18 cut Talen 11 percent and Vistra and Constellation about 4 percent. Sector-wide, merchant power producers are down about 12.5 percent in 2026 while regulated utilities are up roughly 15 percent and renewables 28 percent, a spread Gabelli attributes to "growing concern that the runway for outsized margins could be shortened by government intervention."

And note the institutional echo. The body that triggered October 1929 was a state utility commission, the same species of institution that flipped in Georgia in November 2025 on a ratepayer revolt. In the 1920s that risk detonated all at once. In 2026 it is being amortized through the merchant-power discount. In both eras, though, the deepest damage traced to financial structure rather than headlines: Insull's leverage in 1932, and 2026's sharpest AI drawdowns, Oracle down 32 percent and CoreWeave down 38 percent, coming from capex and counterparty fears rather than politics.

VII. Users over makers: 1927's quiet rotation, 2026's loud one

Inside 1927's 28.75 percent Dow year, a rotation was underway that the index concealed. The companies that made the new technology stumbled, while the companies that used it soared.

The makers hit a wall. By early 1927, sales of radio sets had declined sharply, and the trade press estimated the industry had sacrificed at least $200 million in sales that season, off a 1926 base of roughly $520 million. The first-generation battery-set market had saturated. Sales only recovered in 1928 and 1929 when AC sets forced an upgrade cycle.

The users had the year of their lives. Warner Bros., a struggling studio that bet on Vitaphone sound technology, electricity's killer application for entertainment, premiered The Jazz Singer on October 6, 1927 and watched its stock climb from $9 a share to $132. The Jazz Singer's success pushed Warner earnings 500 percent ahead of 1927, and from 1928 to 1929 profits rose another $12 million at Warner and $3.5 million at Fox as talkies became the norm. The utilities that deployed electrification, not the set-makers that boxed it, were the decade's compounders until the crash.

The lesson of 1927: once the enabling layer saturates, value migrates to the applications built on top of it.

2026 is replaying the rotation, louder and faster. The Magnificent 7, this cycle's set-makers, declined about 1 percent in the first half of 2026 while the S&P 500 rose 9 percent. Meanwhile the users, the real economy adopting and deploying AI, have taken leadership across nearly every breadth measure.

Equal weight over cap weight. The equal-weight S&P 500 is up 14.8 percent against 13.5 percent for the cap-weighted index, the first time it has led since 2022, and it comes after cap-weight's widest three-year winning margin since 1971. Through February it was the strongest relative start since 1992.

Small over large. The Russell 2000 was up 20.1 percent by mid-June against 9.6 percent for the S&P 500, the first sustained small-cap leadership in years.

Value over growth. The Russell 1000 Value ETF was up about 15 percent by mid-June versus roughly 3 percent for its growth counterpart.

Real-economy sectors on top. Energy up 21 percent, materials up 17 percent, staples up 15 percent, and industrials up 12 to 16 percent have led 2026 while the broad market lagged them. The standout industrials are the ones building the AI era's physical layer: GE Vernova and Vertiv, the power-and-cooling equivalent of the theater chains that wired themselves for Vitaphone.

Breadth confirms it. By late February, roughly 65 percent of S&P 500 components were outperforming the index year-to-date, a feat seen only a handful of times in 50 years.

On the 99-year clock, this is 1927 to the decimal point. The enabling-technology makers plateau as their first market saturates, the index keeps climbing anyway, and leadership migrates to the users: studios then, industrials and the other 493 now. In the 1920s version, the makers got one more spectacular act, RCA's 1928 run on the AC upgrade cycle and the Meehan pool, before the political and financial bills arrived together. Whether the 2020s compresses that act too, the way it has compressed everything else, is the live question for 2027.

VIII. The accord 1927 never had: why this decade's endgame won't be 1929–32

Everything above argues the 2020s are rhyming with the 1920s. This section argues where the rhyme breaks: deliberately, structurally, and in our view permanently. The 1920s boom ended the way it did because the system had a brake, and the Federal Reserve stood on it. The 2020s system has had the brake removed. It only has an accelerator.

What the brake did in 1929 to 1932. The Fed of that era sterilized the gold flowing into the United States on the way up, absorbing the world's reserves and neutralizing them, exporting deflation, and then slammed the brake on the way down. When Britain left gold on September 21, 1931 and the U.S. gold stock fell $727 million, or 15 percent, in six weeks, the New York Fed answered the outflow by raising its discount rate from 1.5 percent to 3.5 percent in one week that October, "the sharpest rise within so brief a period in the whole history of the System," in the middle of a depression. Peter Temin calls it one of the most memorable acts of misguided monetary policy in history. In the five months around that decision, 1,860 banks holding $1.45 billion in deposits suspended, and from August 1929 to March 1933 the money stock contracted by more than a third. The Fed defended gold and sacrificed the banking system. That choice, sterilize the inflow and then tighten into the outflow, is what turned a crash into a depression.

Why the choice no longer exists. The Long Island accord of July 1927 was an informal handshake that suspended the brake for a year, and the Fed reversed it by 1928. What we have called the Bessent/Warsh Accord, our name for the observable policy convergence rather than a claimed formal agreement, is the same suspension made structural. Kevin Warsh, confirmed as Fed chair 54 to 45 on May 13, 2026 in the most divisive chair vote in the institution's history, has publicly proposed rewriting the 1951 Treasury-Fed Accord to govern the balance sheet jointly with Treasury: the 1951 separation run in reverse. The operating regime that has emerged pairs standing Fed bill purchases of roughly $360 billion a year with about $180 billion of MBS reinvestment into bills, while Bessent extends dollar plumbing abroad through the FIMA repo channel.

And beneath it sits the GENIUS Act, signed July 18, 2025, which requires every stablecoin dollar to be backed by cash, deposits, or Treasuries of 93 days or less: a structural, non-discretionary bid for the front end that grows with the float. The gold of the 2020s is the tokenized dollar, and the Fed cannot sterilize it. It cannot un-issue a stablecoin, and hiking against the inflow is self-defeating, since higher front-end rates widen issuer carry and accelerate the very deposit flight driving the float. In 1927 to 1931, sterilization was a policy choice. In 2026 to 2028, non-sterilization is baked into the plumbing.

The run that is hard to run. The stress case everyone cites is a stablecoin run, and Warsh has pre-committed to letting one burn: "We do not want to be in the bailout business, full stop." We find a systemic run hard to imagine, though, because of what the float is drawing from. Chinese bank deposits stood at 346 trillion yuan, roughly $48 trillion, at the end of July 2026, still growing 8 percent a year, with Japan adding roughly $8.7 trillion more. Call the accessible low-cost pool $45 trillion and more, conservatively, before any meaningful U.S. or European draw. Chinese households alone hold about $24 trillion in deposits, some 75 to 77 trillion yuan of it in time deposits maturing in 2026 into falling domestic rates, and they are already moving out of deposits in search of yield. A run requires the marginal holder to want out of dollars faster than the next depositor wants in. Against a funding pool that size, the redemption queue refills from the other side. The honest caveat is that this plumbing has never been tested against a genuine float reversal, and Warsh's no-bailout stance means the first test will be run live.

Consequence one: the melt-up should exceed the 1920s. The 1927 to 1929 melt-up happened with the Fed sterilizing most of the inflow. The boom ran against partial resistance, and the Dow still returned 28.75 percent in 1927 and 48.22 percent in 1928. Our framework's expectation, and it is a framework expectation rather than a forecast, is that an unsterilizable inflow produces the same shape at greater amplitude: the 2020s analog years running at roughly double the 1920s percentages, into a 2028 peak on the 99-year clock.

Consequence two: the break comes from abroad, and the bounce holds. On our clock, the 1931-style banking strain maps to bank runs in China and Japan in late 2028. The deposit flight that feeds the float on the way up becomes acute on the way over. We expect that to produce a hard sell-off. But here the 1931 script inverts. In 1931 the crisis abroad forced the Fed to tighten into the crash, because defending gold required it. In 2028, foreign bank stress accelerates flight into tokenized dollars, every one of which must buy bills under GENIUS, so the crisis abroad loosens U.S. financial conditions automatically. The unsterilizable inflow persists through the bust as flight capital. The cushion the 1930 to 1931 Fed refused to allow is now the system's default setting. That is why we expect the post-2028 rebound to be fast and to hold: a V where 1929 to 1932 carved an L.

Consequence three: the 30-year is being retired at the margin. The long bond, the instrument through which markets historically disciplined fiscal policy, is being managed out of the marginal supply. Treasury has held the 30-year auction at $25 billion since May 2024 while bills shoulder the funding load. Its internal projections show 10-, 20-, and 30-year auction sizes being cut from October 2026, and on August 19, 2026 it doubled long-end buybacks to at least $4 billion per operation and doubled their frequency to four per quarter. Roughly 61 percent of dealers surveyed now expect reduced 30-year issuance outright. With the 30-year yield at its highest since 2001, Treasury's answer is not to pay the price but to shrink the market that sets it: issue bills into the GENIUS bid, buy back duration. There is even a Mellon echo. In the 1920s the Treasury retired the debt itself, surplus by surplus. In the 2020s it retires the duration, buyback by buyback. Either way, the bond vigilante's instrument is being decommissioned, which removes the last exogenous brake the 1920s system still had.

The endgame, stated plainly. Same clock, different exit. The 1920s cycle ended in 1929 to 1932 because the monetary system was contractionary by construction: gold sterilized on the way up, rates hiked into the collapse, money supply down a third. The 2020s cycle is expansionary by construction: an inflow that cannot be sterilized, a front end that must be bought, a long bond being retired, and a Fed institutionally merged with the Treasury it once declared independence from.

We expect the melt-up to be larger than 1927 to 1929, the 2028 break to be violent but foreign-sourced, and the recovery to hold, because for the first time since 1913 the central bank's brake pedal has been unbolted from the floor.

Watch the stablecoin float. It is both the fuel gauge and the exit gauge. The month it stalls or reverses is the month this framework starts failing.

JD Unfiltered Research. Analog reasoning, not investment advice. Historical parallels are illustrative frameworks, not forecasts.

The authors hold positions in securities mentioned and reserve the right to buy or sell shares at any time without notice.