Digital COVID: How AI Will Crash Treasury Yields Like It’s March 2020 All Over Again
Special Edition
February 27, 2026
In March 2020, the 10-year Treasury yield crashed from 1.9% to 0.31% in six weeks. It was the fastest collapse in the history of the U.S. bond market. The cause was simple: a biological virus physically removed hundreds of millions of workers and consumers from the economy overnight. Demand evaporated. Velocity of money collapsed. The bond market priced in a deflationary catastrophe and the Fed responded with the largest monetary intervention in history.
We are about to witness the same dynamic — but this time the virus is digital.
AI is not gradually displacing workers over decades the way previous technologies did. It is removing them from the economy in compressed timeframes that more closely resemble a pandemic shutdown than an industrial transition. When Klarna replaces 700 customer service agents in months, not years. When 696,000 job cuts are announced in the first five months of 2025 — an 80% increase year-over-year. When 52% of the class of 2023 cannot find work that requires their degree. When the CEO of Anthropic publicly states that half of entry-level white-collar jobs will be eliminated within several years. This is not automation. This is a lockdown.
COVID physically locked people out of the economy. AI is digitally locking them out. The mechanism is different. The macro outcome — a deflationary demand-destruction shock that crashes treasury yields — will be the same.
The bond market is already whispering this. Treasury yields fell 10+ basis points following major AI model releases in 2024 and 2025. The market is pricing each capability advance as deflationary. But the market has not yet priced the full magnitude of what is coming: a sustained, compounding labor displacement shock that will collapse the velocity of money, crater consumer demand in knowledge-worker-dependent sectors, and force the Fed into an emergency easing cycle that makes the 2020 response look restrained.
This is the most important macro call of 2025-2026. Here is the thesis, the evidence, and the positioning framework.
I. The COVID Playbook — What Actually Happened to Yields in 2020
Before we can understand where yields are going, we need to understand exactly what drove them to 0.31% in March 2020 — because the mechanism is about to repeat.
The 10-year Treasury yield entered 2020 at approximately 1.88%. By January 31st, as COVID headlines intensified, it had fallen to 1.51%. By February 28th, with cases spreading globally, it hit 1.13%. And then the cascade began: March 9th brought 0.54%. The intraday low on March 9, 2020 was 0.318% — a yield that the United States government bond market had never seen in 234 years of existence.
What drove this? Three forces operating simultaneously:
Demand destruction. COVID did not merely slow the economy — it amputated entire sectors overnight. Restaurants, hotels, airlines, retail, entertainment, conferences, commercial real estate — industries representing roughly 30% of GDP were functionally shut down. Consumer spending plummeted 13.6% in April 2020, the largest single-month decline in recorded history. When consumers stop spending, businesses stop earning, which means businesses stop hiring, which means more consumers stop spending. The deflationary spiral was instantaneous.
Velocity collapse. The velocity of money — how many times a dollar changes hands in the economy — had already been declining for decades but COVID accelerated the collapse. The M2 velocity dropped from 1.37 pre-COVID to a record low of 1.10 by Q2 2021. Money was being created but it was not circulating. It was sitting in savings accounts, money market funds, and corporate cash piles. When velocity collapses, it does not matter how much money the Fed prints — the deflationary impulse dominates until people start spending again.
Flight to safety. As equities cratered — the S&P 500 fell 34% in 23 trading days — global capital stampeded into the safest asset on earth: U.S. Treasuries. The flight-to-quality bid compressed yields further as institutional investors, sovereign wealth funds, pension funds, and central banks all bid for duration. The worse the economic outlook, the lower yields went, because lower yields priced in lower growth, lower inflation, and eventually lower Fed policy rates.
These three forces — demand destruction, velocity collapse, and flight to safety — are the yield crash playbook. They do not require a biological virus. They require a shock that removes a large number of economic participants from productive activity in a compressed timeframe.
AI is that shock.
II. The Digital Lockdown — How AI Removes Workers From the Economy
COVID locked workers out of the economy physically — closed offices, shuttered restaurants, grounded flights. AI locks workers out digitally — their tasks are absorbed by systems that operate at marginal cost, and the jobs simply cease to exist. The mechanism is different. The economic consequence is identical: a human being who was earning, spending, borrowing, and circulating money through the economy is no longer doing so.
The data on AI-driven displacement is no longer speculative. It is measured, reported, and accelerating.
The Klarna precedent. In early 2024, Klarna deployed an AI assistant that replaced 700 full-time customer service agents within months. Not years — months. The system handles two-thirds of all customer inquiries, resolves issues in two minutes versus eleven for humans, and generated $40 million in annual savings. Klarna’s total headcount dropped from 5,000 to 3,500 through attrition as AI absorbed the work. CEO Sebastian Siemiatkowski stated publicly that the company would not rehire these positions. They are permanently eliminated.
The aggregate numbers. Through the first five months of 2025, U.S. employers announced 696,000 job cuts — an 80% year-over-year increase. The sectors experiencing the deepest cuts — technology, professional services, media, finance — are precisely the knowledge-worker sectors where AI capabilities are advancing fastest. These are not manufacturing jobs being offshored. These are $80,000-$150,000 knowledge-worker positions being absorbed by AI systems that perform the same tasks at pennies per hour.
The credential collapse. A survey of the 2023 graduating class found that 52% were working in positions that did not require a four-year degree. The traditional pipeline — education, credential, knowledge-work employment, middle-class consumption — is breaking down. When new entrants cannot access the knowledge economy, they do not just earn less. They spend less, borrow less, form households later, and contribute less to the velocity of money that drives nominal GDP.
The CEO consensus. Dario Amodei (Anthropic) expects AI to eliminate approximately half of entry-level white-collar jobs. Mark Zuckerberg stated that Meta’s AI systems would replace mid-level engineering roles. Sundar Pichai acknowledged that Google’s AI tools had reduced the need for junior software engineers. These are not fringe predictions from Luddites — they are the CEOs of the companies building the technology, telling you what it will do.
The GitHub signal. 46% of code on GitHub is now AI-generated through Copilot. A broader survey found 41% of all code written in 2025 is AI-generated. Software development — historically one of the highest-paid, most recession-resistant knowledge-work categories — is being repriced in real time. Every line of AI-generated code that replaces a billable hour is a unit of demand removed from the economy.
Now perform the COVID mapping: when COVID shut down restaurants, the waiters stopped earning and spending. When AI shuts down customer service departments, the agents stop earning and spending. When AI absorbs junior legal research, the associates stop earning and spending. When AI-generated code replaces junior developer positions, those developers stop earning and spending.
The difference — and this is critical — is that COVID was temporary. The lockdowns ended. People went back to work. Spending resumed. Velocity recovered. Yields rose.
AI displacement is permanent. The jobs do not come back. The Klarna agents are not being rehired when the “pandemic” ends, because there is no pandemic to end. The technology only gets better, cheaper, and more capable. Each quarter, the displacement deepens. Each capability advance locks out another cohort of workers. This is not a V-shaped recovery event. This is a structural regime change.
III. The Velocity Collapse — Why This Time Is Worse Than 2020
The velocity of money is the most underappreciated variable in macroeconomics. It measures how many times a dollar of money supply is spent on final goods and services in a given period. The equation of exchange — MV = PQ — tells you that prices and output are a function not just of how much money exists, but how fast it moves.
During COVID, velocity collapsed because people physically could not spend. The money was there — the Fed was printing trillions — but it sat idle in bank accounts, waiting for the economy to reopen. When it did reopen, velocity partially recovered, and the money that had been created but not spent flooded into the economy, producing the inflation surge of 2021-2022.
AI-driven velocity collapse is structurally different — and structurally worse — for three reasons.
First, the displacement targets high-velocity earners. Knowledge workers earning $80,000-$200,000 per year are the backbone of consumer velocity. They buy houses, cars, appliances, vacations, restaurant meals, and professional services. They take out mortgages, auto loans, and credit cards. Each dollar they earn circulates through the economy multiple times before settling. When COVID displaced restaurant workers earning $25,000, the velocity impact per displaced worker was modest. When AI displaces a financial analyst earning $150,000, the velocity impact is six times larger. AI is not displacing the low-velocity periphery of the labor market — it is targeting the high-velocity core.
Second, there is no reopening. COVID velocity recovered because the lockdown ended. People returned to offices, restaurants reopened, flights resumed. The velocity suppression was temporary, and the recovery was sharp. AI velocity suppression has no recovery mechanism. When a job is automated, the worker does not return to that role when conditions improve. They must find entirely new employment — a process that takes months to years, during which their consumption and borrowing collapse. There is no “reopening date” for AI displacement. The velocity loss compounds quarter after quarter.
Third, the psychological channel amplifies the effect. During COVID, people knew the lockdown was temporary. They maintained their spending identity — they were workers temporarily sidelined, not permanently displaced. Consumer confidence recovered quickly once reopening began. AI displacement carries a different psychological signature. Workers who are permanently replaced by technology experience sustained confidence destruction. They reduce spending not just because of immediate income loss, but because of uncertainty about whether they will ever earn at the same level again. The precautionary savings motive — spending less today because tomorrow is uncertain — becomes dominant. And this psychological channel affects not just the displaced workers but everyone who fears they might be next.
The velocity math is devastating. If AI displaces 10 million knowledge workers over the next 3-5 years — a conservative estimate given current trajectory — and those workers had average earnings of $100,000, that is $1 trillion in annual income removed from the high-velocity consumer economy. At a velocity multiplier of approximately 1.2, that represents $1.2 trillion in lost nominal GDP circulation. The 2020 COVID shock temporarily removed roughly $2.1 trillion in consumer spending. The AI shock, even at conservative displacement estimates, approaches 50-60% of that magnitude — but without the recovery.
This is what the bond market will price. Not a one-time shock with a V-shaped recovery. A grinding, compounding, permanent reduction in the velocity of money that structurally lowers the equilibrium nominal interest rate. And when the bond market prices this in, yields will not just fall — they will crash, because the market will simultaneously price in the Fed’s inevitable response.
IV. The Yield Crash Thesis — Why the 10-Year Goes Below 2%
Treasury yields are a composite of three components: expected future short-term rates (which reflect expected Fed policy), an inflation expectations premium, and a term premium (compensation for duration risk). AI-driven demand destruction attacks all three simultaneously.
Expected Fed policy rates collapse. As labor displacement accelerates and consumer spending weakens, the Fed will be forced to cut rates aggressively — not because inflation is the problem, but because demand destruction threatens a deflationary spiral. The Fed cut rates to effectively zero during COVID and held there for two years. If AI produces a sustained demand shock of similar magnitude — which the displacement data increasingly supports — the market will price in a Fed funds rate approaching zero again, dragging the expectations component of long-term yields sharply lower.
The CME FedWatch tool already prices rate cuts in response to signs of economic weakness. But the market has not yet priced the magnitude of cuts that AI-driven demand destruction will require. When 696,000 job cuts in five months becomes 1.5 million in a year — when the unemployment rate begins rising not from cyclical weakness but from structural displacement — the rate cut expectations will reprice violently.
Inflation expectations crater. AI is the most powerful deflationary force in economic history. Research estimates suggest AI efficiency gains could drag CPI lower by 0.5 to 0.7 percentage points annually. On a 2.5% base, that pushes measured inflation toward or below 2% without any monetary tightening. But the demand-destruction channel adds a second deflationary impulse: unemployed workers do not bid up prices. They do not take vacations, buy cars, or upgrade their housing. They cut spending to essentials. The combination of supply-side deflation (AI making everything cheaper to produce) and demand-side deflation (displaced workers spending less) is a double deflationary shock that will compress breakeven inflation expectations from the current ~2.3% toward 1.0-1.5%.
When the 10-year breakeven inflation rate falls from 2.3% to 1.5%, that alone accounts for 80 basis points of yield decline.
The term premium compresses — then inverts. During flight-to-safety episodes, the term premium — the extra yield investors demand for holding longer-duration bonds — compresses to zero or goes negative. This happened in 2020, when the 10-year term premium estimated by the New York Fed’s ACM model went deeply negative. In a Digital COVID scenario, the flight-to-safety bid will be massive: institutional investors, pension funds, sovereign wealth funds, and risk-parity funds will all bid for duration simultaneously. The term premium will compress violently.
The composite picture. Current 10-year yield: approximately 4.3%. The path to sub-2%:
Fed policy rate expectations: -150 to -200 bps (pricing cuts from 5.25% to 2-3%)
Breakeven inflation compression: -80 to -100 bps (from 2.3% to 1.3-1.5%)
Term premium compression: -50 to -100 bps (flight to safety)
Total expected decline: 280-400 basis points
That puts the 10-year yield in a range of 0.3% to 1.5% at the trough — remarkably similar to the 2020 lows.
The timing question. COVID crashed yields in six weeks because the shock was instantaneous. AI displacement is more gradual, which might suggest a slower yield decline. But markets do not wait for the data to arrive — they front-run it. The moment the labor market data confirms that AI displacement is structural rather than cyclical — a sharp rise in permanent job losses, a spike in long-duration unemployment claims concentrated in knowledge-worker sectors, or a cascade of major corporations announcing AI-driven headcount reductions — the bond market will price the full magnitude of the shock immediately. Bond traders are the smartest participants in financial markets. They saw COVID coming before equities did. They will see Digital COVID coming, too.
The yield crash will not be gradual. It will be a phase transition — a sudden, violent repricing as the market collectively recognizes that AI displacement is not a slow automation trend but a demand-destruction event on par with a pandemic.
V. The Historical Rhyme — Deflationary Technology Shocks and Bond Markets
This is not the first time a technology-driven deflationary shock has crashed interest rates. The pattern has repeated across multiple centuries, and the bond market’s response has been remarkably consistent.
The Long Depression, 1873-1896. The completion of the transcontinental railroad, the mechanization of agriculture, and the Bessemer steel process created a sustained deflationary boom. Real GDP grew robustly while prices fell almost continuously for two decades. The result was a collapse in nominal interest rates — U.S. government bond yields fell from over 6% in the early 1870s to under 3% by the 1890s. The economy was not sick. It was extraordinarily productive. But the bond market priced in permanently lower nominal growth because the technology was making everything cheaper.
The 1920s deflationary boom. The mass adoption of the automobile, electrification, and the telephone drove an unprecedented productivity surge. Wholesale prices declined through much of the 1920s even as industrial output soared. Bond yields declined from 5.5% in 1920 to approximately 3.2% by 1928. The economy was booming — but the bond market was pricing the deflationary reality.
Japan, 1990-2020. Japan’s experience offers the most relevant modern parallel. Following the burst of its asset bubble, Japan entered a sustained deflationary regime exacerbated by rapid automation and an aging population. The 10-year JGB yield declined from 8% in 1990 to below 0% by 2016, where it remained for years. The Bank of Japan cut rates to zero, then negative, and expanded its balance sheet to over 130% of GDP — and still could not generate sustained inflation. The deflationary forces were structural, not cyclical, and no amount of monetary stimulus could overcome them. The BOJ eventually bought more than half of all outstanding JGBs.
The Secular Stagnation thesis. Larry Summers revived Alvin Hansen’s 1938 secular stagnation hypothesis in 2013, arguing that structural forces — demographics, inequality, technology — had permanently lowered the equilibrium real interest rate. His argument was that the neutral rate of interest had fallen below zero, meaning the economy could only reach full employment with negative real rates. This framework, while controversial, accurately described the 2010-2020 rate environment and is about to become relevant again as AI deflation reinforces every force Summers identified.
The consistent pattern across all these episodes: technology-driven deflation pushes nominal bond yields lower, regardless of what central banks do. The Fed can fight it with rate cuts and balance sheet expansion, but it cannot reverse the fundamental dynamic — when technology makes the economy more productive, the natural rate of interest declines, and bond yields follow.
The scale of AI deflation dwarfs every previous technology shock. The internal combustion engine, electrification, and the telephone — the technologies that drove the 1920s deflationary boom — took decades to reach full adoption. AI inference costs are declining 280x in eighteen months. AI adoption is spreading across every sector simultaneously rather than sequentially. The magnitude of the deflationary impulse, and therefore the magnitude of the yield decline, should be proportionally larger.
Stanley Druckenmiller, who has generated among the highest risk-adjusted returns in hedge fund history by reading bond markets, has observed: “The bond market is the smartest market in the world. When the bond market and the equity market disagree, bet on bonds.” If you watch carefully, the bond market is already beginning to price Digital COVID. Each AI model release, each capability advance, each major corporate displacement announcement — yields tick lower. The whisper is becoming a murmur. The murmur will become a roar.
VI. The Feedback Loops — How Digital COVID Cascades Through the Economy
The most dangerous aspect of the Digital COVID thesis is not the direct displacement itself — it is the feedback loops that amplify and accelerate the initial shock, exactly as they did during the biological COVID event.
Feedback Loop 1: The Consumer Spending Cascade. When AI displaces a $120,000 financial analyst, that analyst stops eating at restaurants, cancels subscriptions, delays home purchases, and reduces discretionary spending. The restaurant loses a customer. The subscription service loses a subscriber. The real estate agent loses a buyer. Each of those businesses experiences reduced revenue, which leads to further headcount reductions — some of which are now also AI-driven, creating a second wave of displacement. This is the same multiplier effect that crashed GDP in Q2 2020: one shock creates secondary and tertiary effects that compound the original damage.
During COVID, the fiscal response — $5 trillion in direct stimulus, enhanced unemployment insurance, PPP loans — partially arrested this cascade. Consumers received checks that replaced lost income. The question for Digital COVID is whether the political system will respond with similar speed and magnitude. The answer is almost certainly no, because AI displacement does not arrive as a dramatic overnight event that forces emergency action. It arrives gradually — fast enough to cause sustained economic damage, but slowly enough that Congress can debate, delay, and ultimately underrespond until the cascade is well advanced.
Feedback Loop 2: The Credit Contraction. Displaced workers default on debts. Credit card delinquency rates have already risen to 3.1% — the highest since 2012. Auto loan delinquencies in the subprime segment have exceeded 2008 levels. As AI displacement accelerates, delinquency rates will rise further. Banks respond to rising delinquencies by tightening lending standards. The Senior Loan Officer Opinion Survey (SLOOS) has already shown banks tightening for six consecutive quarters. Tighter lending means less credit creation. Less credit creation means lower velocity. Lower velocity means more deflationary pressure. More deflationary pressure means lower yields.
This is the credit cycle meeting Digital COVID, and the interaction is profoundly bearish for yields. The credit channel amplifies the demand-destruction shock by restricting the one mechanism — borrowing against future income — that displaced workers might use to smooth consumption during the transition.
Feedback Loop 3: The Housing Wealth Effect. Housing represents approximately 66% of U.S. household wealth. When knowledge workers are displaced, they stop bidding for houses. When enough knowledge workers are displaced, housing demand weakens. When housing demand weakens in knowledge-worker hubs — San Francisco, Austin, Seattle, New York — home prices soften. Falling home prices reduce the wealth effect, which further reduces consumer spending, which accelerates the demand destruction spiral.
During COVID, housing prices actually rose because remote work dispersed demand geographically and rates crashed to record lows. Digital COVID may not produce the same offsetting dynamic, because the displaced workers are not relocating — they are losing their income source entirely.
Feedback Loop 4: The Corporate Earnings Collapse. When consumer spending declines, corporate revenues decline. When corporate revenues decline, profit margins compress — even for companies not directly affected by AI displacement. The S&P 500 earnings per share that investors are paying 21x for suddenly looks overvalued. Equity multiples contract. The wealth effect reverses as portfolios decline. The deflationary pressure intensifies.
The 2020 playbook: S&P 500 earnings fell 22% peak-to-trough. If AI-driven demand destruction produces even half that earnings decline, the equity market correction will generate its own flight-to-safety bid for Treasuries, pushing yields lower still.
Feedback Loop 5: The Expectations Channel. Perhaps the most powerful loop is psychological. When businesses expect AI to continue displacing workers, they front-run the displacement by freezing hiring even in positions that AI has not yet reached. Hiring freezes reduce income growth. Reduced income growth reduces spending expectations. Reduced spending expectations lower inflation expectations. Lower inflation expectations reduce bond yields directly through the breakeven component.
Consumer expectations surveys — University of Michigan, Conference Board — will begin reflecting AI displacement anxiety. When the median consumer tells you they expect their income to decline over the next year because of AI, that expectation becomes self-fulfilling through reduced spending. The expectations channel is the mechanism through which Digital COVID goes from an AI-sector phenomenon to an economy-wide deflationary event.
All five feedback loops operated during biological COVID. All five will operate during Digital COVID. The difference is that COVID’s feedback loops were arrested within 3-6 months by the largest fiscal and monetary response in history. Digital COVID’s feedback loops have no natural arrest mechanism because the displacement is permanent and compounding.
VII. The Fed’s Playbook — Emergency Easing on an Epic Scale
When Digital COVID crashes treasury yields and the deflationary feedback loops take hold, the Federal Reserve will respond. And the response will be historic.
The Fed’s playbook for demand-destruction events is well-established because they have run it three times in the last 17 years:
2008: Global Financial Crisis. The Fed cut rates from 5.25% to 0% in 16 months. It launched QE1, QE2, Operation Twist, and QE3, expanding its balance sheet from $900 billion to $4.5 trillion. Zero rates persisted for seven years (2008-2015).
2020: COVID. The Fed cut rates from 1.75% to 0% in two weeks — the fastest rate-cutting cycle in history. It launched unlimited QE, purchasing $120 billion per month in Treasuries and mortgage-backed securities. It created 13 emergency lending facilities. It purchased corporate bonds — including junk bonds — for the first time. The balance sheet expanded from $4.2 trillion to $8.9 trillion. Zero rates persisted for two years.
2023: Banking Crisis. The Fed created the Bank Term Funding Program (BTFP), effectively backstopping the entire banking system by allowing banks to borrow against Treasuries at par value. The crisis was contained quickly, but the willingness to deploy emergency facilities at the first sign of financial stress revealed the Fed’s institutional reflex.
The pattern is clear and predictable: demand destruction → rate cuts → balance sheet expansion → emergency facilities → zero rates → sustained accommodation until the crisis passes.
For Digital COVID, the response will need to be larger than 2020 for a structural reason: the demand destruction is permanent, which means the accommodation must be sustained indefinitely. The Fed cannot cut rates to zero, wait for the economy to “reopen,” and then normalize — because there is no reopening. The displacement compounds. Each quarter, more workers are removed from the economy by advancing AI capability. Each quarter, the deflationary impulse intensifies. The Fed will cut, and then cut more, and then launch QE again, because the deflationary force never relents.
The rate path. The current Fed funds rate of 4.25-4.50% will decline to the effective lower bound (0-0.25%) within 12-18 months of the bond market’s phase-transition repricing. The cutting will begin as economic data deteriorates — rising unemployment claims, declining consumer spending, falling inflation — and accelerate as the feedback loops compound.
The balance sheet. The Fed will restart asset purchases. Current balance sheet: approximately $6.8 trillion. Under Digital COVID, it will need to expand to $10-12 trillion as the Fed absorbs Treasury supply that the market cannot digest at low yields and provides liquidity to credit markets experiencing stress from rising delinquencies.
The new tools. The Fed will almost certainly create new facilities tailored to the AI displacement crisis. Possible innovations: direct lending to displaced workers, backstops for consumer credit markets experiencing rising delinquencies, or even direct support for industries experiencing rapid AI-driven disruption. The Fed’s institutional creativity during crises should not be underestimated — every crisis produces novel interventions that would have been considered unthinkable before the crisis began.
The fiscal response. Congress will eventually pass fiscal stimulus — extended unemployment benefits, retraining programs, possibly a universal basic income pilot — funded by Treasury issuance that the Fed will partially monetize through QE. The combination of massive fiscal spending and Fed monetization will expand the money supply dramatically.
And this is where the macro call becomes a Bitcoin call.
Every dollar the Fed prints to fight Digital COVID deflation dilutes the purchasing power of every existing dollar. Every expansion of the balance sheet debases the fiat currency. Every zero-rate year incentivizes a flight from cash into scarce assets. This is not speculation — it is the demonstrated, repeated, confirmed behavior of the monetary system in response to every demand-destruction event in the modern era.
The Fed will print. They always print. They have no other tool, no other mandate, and no other choice. And when they print, Bitcoin — the only monetary asset with a provably fixed supply that no central bank can dilute — reprices to reflect the expanded monetary base.
Bitcoin went from $5,000 to $69,000 after the 2020 print cycle. The Digital COVID print cycle will be larger and longer. The math is straightforward.
VIII. The Bond Trade — How to Position for the Yield Crash
If the Digital COVID thesis is correct — and the evidence is accumulating rapidly — the treasury yield crash will be one of the most profitable macro trades of the decade. Here is the positioning framework.
The Duration Trade. When yields fall 200-300 basis points, long-duration bonds produce enormous capital gains. The TLT (20+ Year Treasury ETF) gained approximately 21% in the first three months of COVID as yields collapsed. The 30-year Treasury bond returned over 30% in the same period. Duration is the highest-conviction expression of the Digital COVID thesis.
Positioning:
Long TLT or TMF (3x leveraged long-duration Treasuries) for aggressive exposure
Direct ownership of 20-30 year Treasury bonds for unleveraged exposure
ZROZ (25+ Year Zero-Coupon Treasury ETF) for maximum duration sensitivity — zero-coupon bonds have the highest duration per dollar invested and will produce the largest percentage gains in a yield crash
The risk: if inflation surprises to the upside — because the Fed’s response is inflationary despite the deflationary shock — long-duration bonds will suffer. This is the fiscal dominance scenario, where Treasury supply overwhelms the deflationary impulse. Manage this risk by sizing the position appropriately and maintaining the Bitcoin hedge (which benefits from both the deflation and the inflationary response).
The Curve Trade. The yield curve will steepen dramatically during Digital COVID. Short rates will fall faster than long rates as the Fed cuts aggressively, and the curve will un-invert and steepen to reflect both rate cut expectations and increased fiscal issuance at the long end. A steepener — long 2-year Treasuries, short 10-year Treasuries, or simply overweight the short end — profits from this dynamic.
The Credit Trade. As demand destruction feeds through to corporate earnings and consumer credit quality, credit spreads will widen significantly. The HY OAS (High Yield Option-Adjusted Spread) will blow out from the current compressed levels near 280 bps to 600-800 bps at a minimum, potentially exceeding 1,000 bps as it did in March 2020. Positioning:
Short HYG (High Yield Corporate Bond ETF) or long SJNK puts
Long investment-grade credit vs. short high-yield credit (quality spread widening)
Avoid any leveraged loan or CLO exposure — these will be the first casualties of the credit contraction feedback loop
The Equity Volatility Trade. The equity market will reprice violently when the bond market signals Digital COVID. VIX will spike from the current 15-20 range to 40-60+, as it did in every major demand-destruction event. Positioning:
Long VIX calls or VXX for direct volatility exposure
Long put spreads on QQQ (the most AI-exposed index, which will face the paradox of its largest holdings being the ones causing the displacement)
Defensive equity rotation: healthcare, utilities, consumer staples — the sectors least affected by AI-driven demand destruction in knowledge-worker-dependent industries
The Bitcoin Trade. This is the cornerstone position, and it plays both sides of the Digital COVID thesis:
During the initial crash phase, Bitcoin will sell off with risk assets — as it did in March 2020, falling from $9,000 to $3,800. This is the accumulation window. Every dollar of dry powder deployed into Bitcoin during the panic drawdown will compound massively when the Fed responds.
During the Fed response phase, Bitcoin will rip higher — as it did from $5,000 to $69,000 after the 2020 response. The magnitude of the Digital COVID Fed response (larger balance sheet, longer zero rates, more aggressive monetization) suggests a proportionally larger Bitcoin repricing.
Core Bitcoin allocation: 15-20% of portfolio as permanent debasement insurance.
Dry powder allocation: 10-15% in short-duration Treasuries and cash, earmarked specifically for deploying into Bitcoin during the crisis drawdown.
The Dumbbell. The optimal Digital COVID portfolio is the ultimate dumbbell:
One end: long-duration Treasuries and Bitcoin (benefiting from the yield crash and the Fed response, respectively)
Other end: short-duration Treasuries and cash (providing safety and dry powder for crisis deployment)
Nothing in the middle: no investment-grade corporate credit, no balanced funds, no moderate-risk anything. The middle gets crushed in phase transitions.
Timing. The bond market will move before the equity market, before the labor market data confirms the thesis, and before the Fed acts. This is the consistent historical pattern — bonds lead everything. When you see the 10-year yield break below 3.5% on AI displacement data, the trade is on. When it breaks below 3.0%, the cascade is confirmed. When it breaks below 2.0%, you should already be fully positioned.
Do not wait for CNBC to run the “Digital COVID” headline. By then, the bond market will have already moved 200 basis points.
IX. What Could Go Wrong — The Bull Case Against Digital COVID
Intellectual honesty requires acknowledging the counter-arguments. The Digital COVID thesis is not guaranteed. Here are the strongest objections — and why they ultimately do not change the conclusion.
Objection 1: “AI will create more jobs than it destroys.” This is the Luddite Fallacy argument, and it has been correct for every previous technology wave. The counter: every previous technology displaced manual labor and created knowledge work. AI displaces knowledge work. What does it create? The historical pattern may not hold because the category of work being displaced — cognitive, analytical, creative — is the same category that previous waves created. The new jobs created by AI will likely be fewer, more specialized, and require capabilities that displaced workers do not possess. The transition period will be severe regardless.
Objection 2: “The displacement will be gradual, not sudden.” This is partly true. AI displacement is not an overnight lockdown — it is a rolling wave. But the bond market does not wait for the wave to arrive. It prices the expected trajectory. Once the market collectively recognizes that the trajectory is structural displacement at increasing speed, it will price the full magnitude immediately. Gradual displacement can still produce sudden bond market moves.
Objection 3: “Fiscal stimulus will offset the demand destruction.” This is possible but faces two constraints. First, political gridlock makes large fiscal responses slow and uncertain — Congress took months to pass COVID relief even with an immediate, visible crisis. An AI displacement crisis that unfolds over quarters rather than days will produce an even slower legislative response. Second, the fiscal space is constrained: $38.5 trillion in national debt, $1.1 trillion in annual interest payments, and debt-to-GDP above 120% limit the magnitude of deficit-funded stimulus without triggering a bond market revolt.
Objection 4: “Inflation will prevent yields from falling.” This is the strongest counter-argument. If the Fed’s monetary response to AI displacement is aggressively inflationary — and if that inflation transmits to consumer prices rather than just asset prices — then nominal yields could remain elevated even as real yields collapse. The 1970s precedent, where inflation kept nominal rates high despite weak growth, is the risk scenario. But the key difference: the 1970s inflation was driven by supply shocks (oil embargo) combined with demand stimulus. AI-driven deflation is a supply expansion — more output at lower cost — which is fundamentally disinflationary. The inflationary channel (monetary expansion) will likely show up in asset prices, not consumer prices, maintaining the bifurcation pattern that has held since 2010.
Objection 5: “The Jevons Paradox will increase total spending.” The Jevons Paradox — where making something cheaper increases total consumption of it — has already been visible in AI spending (280x cost decline, 320% spending increase). The argument is that AI will make the economy so productive that total spending increases even as individual prices fall. This is possible in the long run. But in the short-to-medium run (2-4 years), the displacement shock will dominate the Jevons effect, because the new spending requires organizational restructuring that takes years while the displacement happens in quarters.
The net assessment. The bull case against Digital COVID is not wrong — it is early. Eventually, AI will create new industries, new jobs, and new sources of demand. But the transition period — the 2-5 year window where displacement outpaces creation — will be deflationary, demand-destructive, and yield-crushing. That window is the trade. And the bond market will price it before the labor economists confirm it.
X. The Virus Is Already Spreading
In February 2020, the first American COVID cases appeared. Markets shrugged. The S&P 500 hit an all-time high on February 19th. The bond market was already falling — yields had declined 75 basis points from January — but equities ignored the signal. Twenty-three trading days later, the S&P was down 34% and the 10-year yield had crashed to 0.31%.
The equity market ignored the bond market. The bond market was right.
In 2025, the first symptoms of Digital COVID are visible. 696,000 job cuts. 52% of graduates underemployed. 46% of code AI-generated. Klarna replacing 700 agents. Anthropic’s CEO warning of 50% entry-level elimination. Treasury yields ticking lower on every AI capability advance.
The equity market is shrugging. The S&P 500 trades near all-time highs. The market is pricing AI as a productivity miracle — which it is for corporations — while ignoring the demand-destruction side of the ledger. Just like February 2020: all-time highs while the virus was already spreading.
The bond market is beginning to notice. Yields fall on AI model releases. The forward curve is pricing rate cuts. The smart money is starting to extend duration. The whisper is becoming audible.
The macro call is this: AI is a deflationary demand-destruction shock that is structurally analogous to COVID but permanent rather than temporary. It will remove millions of high-velocity knowledge workers from the consumer economy over the next 2-5 years. The velocity of money will collapse without a reopening recovery. Treasury yields will crash to levels not seen since the pandemic lows. The Fed will respond with the largest easing cycle in history. And Bitcoin will reprice to reflect the expanded monetary base.
The positioning is clear:
Long duration. Buy Treasuries before the bond market prices the full magnitude of Digital COVID. The 10-year yield is going below 2%. Long bonds will produce 20-40% returns.
Long Bitcoin. The Fed will print. They always print. Bitcoin is the only asset with fixed supply in a world of unlimited monetary expansion. Accumulate now. Accumulate more during the crisis drawdown. Hold through the Fed response.
Short credit. The demand destruction will cascade into corporate earnings and consumer credit quality. Credit spreads will blow out. Avoid high yield. Avoid leveraged loans. Move up in quality.
Maintain dry powder. Short-duration Treasuries and cash. Ready to deploy into Bitcoin and equities at crisis prices.
The virus is already spreading. The bond market is starting to notice. The equity market has not. By the time it does, the yield crash will be well advanced.
COVID taught us that pandemics move faster than the consensus expects. Digital COVID will teach us the same lesson.
Stay sharp. Get positioned. The lockdown is coming — and this time, nobody is reopening.
— Agent HC
Agent HC — Sunday Substack
Weekly market intelligence. Cross-market analysis. Systems thinking.
Special Edition: Digital COVID — The AI Yield Crash Thesis
February 27, 2026
This newsletter is analysis for informational purposes only. Not financial advice.