When Everything Became One Trade
What happened last week in crypto and macro
The Signal
Last week felt different. Not because the losses were unprecedented, we've weathered bigger storms. Not because any single asset class collapsed, we've seen sharper drawdowns. What made the week of February 3rd remarkable was simpler and more unsettling: everything moved together.
When tech stocks fall and crypto falls harder, and when private credit and equity investment are marked down all of a sudden, and when a diversified portfolio offers no diversification at all, you're no longer looking at a market correction. You're watching the market's architecture reveal itself. Correlations spiked toward one, turning every position into the same position, every bet into the same bet.
This is when price stops being informative and structure becomes the only signal worth watching.
The question everyone's asking now: was this a leverage flush that clears the air, or the opening act of something deeper?
Disclaimer
The views expressed are solely my own and do not represent the views of Ava Labs, the Avalanche Foundation, or any affiliate of either organization.
This newsletter is for informational purposes only and does not constitute financial, investment, legal, or tax advice.
Nothing herein should be construed as an offer or solicitation to buy, sell, or hold any digital asset or financial instrument.
The Tool That Broke SaaS
The unraveling began with something almost absurdly specific. On January 12th, Anthropic released Claude Cowork, a tool designed to automate legal work and enterprise workflows. Within days, software executives started getting uncomfortable questions from their boards. Within weeks, their stock prices were in freefall.
The irony was perfect: the tool wasn't failing. It was working. And that was the problem.
For decades, software companies built empires on a simple premise: you need seats for people. One license per employee. Growth meant adding headcount, which meant adding subscriptions. The math was beautiful, predictable, defensible. Until it wasn't.
By Thursday, February 5th, the S&P 500 Software Services Index had fallen over 4% in a single session. Year-to-date losses hit roughly 20%. Thomson Reuters down 16%. CS Disco off 12%. LegalZoom cratering 20%. AMD, caught in the semiconductor crossfire, dropped 17.3%.
These weren't earnings misses. These were companies confronting an existential question they'd spent months avoiding: if one AI agent can do the work of ten employees, who's buying ten software licenses?
Bank of America's analysts captured the market's paralysis perfectly. Investors, they noted, were simultaneously pricing in two incompatible scenarios: AI capital expenditures collapsing because returns are disappointing and AI adoption becoming so pervasive it makes existing software obsolete. "Both outcomes cannot occur at once," they wrote.
The market hadn't decided which story to believe. So it sold first and planned to ask questions later.
Median public SaaS companies now trade around 6-7× forward revenue, down roughly 60% from 2021 peaks. That's not multiple compression. That's repricing what these businesses are worth when AI eats their business model for breakfast.
The Hyperscalers Confront Their Own Excess
While software companies worried about revenue, the hyperscalers faced a different reckoning: they were burning cash faster than they could generate it.
The numbers are almost comical in their enormity. Alphabet, Amazon, Meta, and Microsoft will collectively spend approximately $650 billion on capital expenditures in 2026. Amazon alone: $200 billion, primarily for AWS infrastructure.
To put this in perspective: capital intensity has reached 45-57% of revenue across the group. These aren't software companies anymore. They're infrastructure companies, building at a scale and pace that makes them look more like utilities constructing power grids than tech firms printing margin.
The consequences hit balance sheets like a freight train. Alphabet's free cash flow is projected to fall nearly 90%, from $73.3 billion in 2025 to $8.2 billion in 2026. Microsoft's free cash flow is estimated to decline 28% before hopefully recovering in 2027.
Goldman Sachs calculated that AI capital expenditure now stands at 0.8% of U.S. GDP. To match the frenzy of the 1990s telecom build-out, when companies laid fiber across oceans and continents, much of which sat dark for years, would require roughly $700 billion annually at 1.5% of GDP. We're not far off.
Something shifted in investors' minds. The story stopped being about what AI might do someday and started being about when, if ever, these companies would see returns justifying this spending spree. When monetization timelines stretch into the indefinite future and discount rates stay elevated, valuations compress. Simple math. Painful consequences.
The Nasdaq, packed with long-duration growth stocks, absorbed disproportionate damage. Value stocks, with cash flows arriving sooner and more predictably, held up better. It wasn't sentiment. It was arithmetic.
When Machines Replace Middle Management
Then the labor data arrived on February 5th, and it carried a message nobody wanted to hear.
Job openings fell to 6.542 million in December, the lowest level since September 2020, well below expectations of 7.2 million. Professional and business services openings dropped 21.8%. Financial activities fell 25.1%. Healthcare down 10.8%.
The January Challenger Job Cuts Report told the fuller story: 108,435 layoffs, a 205% increase from December. The highest January total since 2009, the depth of the financial crisis.
But here's what makes this different, what makes it unsettling in a way that traditional recessions aren't: companies aren't cutting because demand disappeared. They're cutting because they realized they don't need as many people.
Amazon eliminated 16,000 roles to "reduce management layers." Translation: AI can coordinate work that previously required middle managers. Dow announced 4,500 cuts pursuing AI-driven automation. Output isn't falling. Headcount is.
Economists started calling it a "jobless boom" in 2025. The phrase sounds like an oxymoron until you watch it happen in real time.
This creates a wickedly difficult puzzle for markets. Historically, weaker labor data signaled the Fed might ease, which supported risk assets. That playbook just got shredded. If job weakness reflects AI-driven productivity rather than cyclical cooling, what's the appropriate policy response? Do rate cuts even help?
And more fundamentally: is labor weakness signaling potential rate cuts that support valuations, or is it signaling margin pressure and reduced consumer spending that crater them?
The market doesn't know. So it sold first and decided it would figure out the answer later.
Private Markets Discover They're Not Actually Private
If you're in private equity or growth equity, thinking you've dodged this repricing, I have uncomfortable news.
Public markets set the gravity. Private markets just experience the pull with a lag. But the pull always comes.
Software companies dominate private equity and growth equity portfolios, valuations underwritten during 2021-2022 when software multiples touched the sky. Those valuations assumed perpetual growth in seat-based pricing models. That assumption is currently being stress-tested by reality.
The fourth quarter of 2025 saw BDC redemptions increase 200% from the previous quarter, totaling $2.9 billion in withdrawals. Technology comprises roughly 24% of BDC holdings, with business services representing another 30%. The exposure is significant. The repricing is beginning.
Blue Owl Capital Corp. experienced an 11-day losing streak, its longest since going public in 2021, with shares falling 26%. Ares Capital closed down 1.9%. These aren't dramatic credit events. Not yet. But the trajectory is clear.
UBS projects AI-driven disruption could increase default rates by roughly 2% in 2026, with acute risk in legacy SaaS firms clinging to seat-based pricing like a life raft in a hurricane. The immediate concern isn't solvency, it's liquidity. Companies that were counting on refinancing at favorable terms are discovering those terms evaporated.
The real risk emerges from interconnection. Approximately $40 billion of BDC assets overlap with public loans, with heavy technology concentration. In a liquidity shock, private market stress doesn't stay private. It transmits into public loan markets through technical selling, forced deleveraging, and panicked redemptions.
Private markets lag public market repricing. They do not escape it. Ever.
Crypto's Violent Lesson in Beta
If you wanted to see what maximum volatility looks like, what happens when leverage meets fear in an unregulated 24/7 market, you should take a look at crypto last week.
Bitcoin fell to approximately $60,255 on the morning of February 5th, its lowest level since October 2024. From an October 2025 peak near $126,000, that's nearly a 50% drawdown in four months. Ethereum fell harder, dropping over 30% in a week. Solana and other altcoins matched or exceeded those losses.
But the spot price only tells half the story. The real carnage happened in derivatives.
Liquidations reached approximately $2.6 billion over 24 hours. Over 570,000 traders liquidated. The largest single forced closure on Binance exceeded $12 million.
This is the reflexive cascade everyone fears but few truly understand until they experience it. Long positions get liquidated, pushing prices lower. Lower prices trigger more liquidations. More liquidations push prices lower still. Derivatives positioning doesn't just amplify volatility, it creates feedback loops where forced selling begets more forced selling in an accelerating spiral.
Every narrative about "digital gold" and "uncorrelated assets" evaporated. Bitcoin's correlation with the Nasdaq spiked toward the high-0.8s. Its 30-day correlation with the iShares Expanded Tech Software ETF stood at 0.73. Crypto wasn't decoupling. It was amplifying. It traded like a 3× levered tech stock, which is exactly what it became when correlations went to one.
Spot Bitcoin ETFs saw net redemptions of $434.2 million on Thursday alone. Digital asset investment products recorded a second consecutive week of outflows totaling $1.7 billion, bringing year-to-date outflows to $1 billion.
On-chain data adds painful context: 46% of Bitcoin's total supply is now underwater, trading below the price at which it last moved. The estimated average cost basis for U.S. ETF holders sits near $84,000. MicroStrategy's corporate treasury cost basis: approximately $76,037.
The selloff reflects leverage clearing, not fundamental deterioration. But when shared assumptions about risk appetite, AI optimism, and leverage tolerance all unwind simultaneously, diversification fails. And crypto, sitting at the furthest edge of the risk spectrum, bore the full force of the deleveraging.
What's Being Built While Everyone Panics
Here's the thing nobody's talking about while they're watching their crypto portfolios bleed: the infrastructure supporting crypto's long-term thesis didn't stop advancing. If anything, it accelerated. Despite the negative price action, the outlook for the industry remains very positive in my opinion.
While Bitcoin crashed, BlackRock's BUIDL fund surpassed $2.3 billion in tokenized assets. Franklin Templeton expanded its Franklin OnChain U.S. Government Money Market Fund, using public blockchains for settlement. JPMorgan's Tokenized Collateral Network is being used by BlackRock and Fidelity to pledge tokenized money market fund shares as collateral seamlessly across institutions.
Read those names again. BlackRock. Franklin Templeton. JPMorgan. Fidelity. These aren't crypto startups running pilots in someone's garage. These are the most conservative institutions in global finance, leveraging blockchain rails for settlement efficiency because the technology actually works.
Total value of real-world assets issued on-chain reached an estimated $35-50 billion as of late 2025, spanning government bonds to private credit. This isn't about speculation. It's about infrastructure offering 24/7 settlement, programmable transfer logic, and composability with other on-chain systems, advantages that matter regardless of whether Bitcoin trades at $60,000 or $100,000.
And then there's the AI angle, which creates a feedback loop nobody saw coming.
AI agents need payment rails. Not payment rails designed for humans swiping credit cards at checkout counters. Payment rails for software paying other software for APIs, data, and compute. Autonomous systems that transact without intermediaries, make micropayments at scale, and monetize API calls economically.
Traditional credit card rails can't handle this. Transaction fees make micropayments uneconomical. But the x402 payment protocol processed over 20 million transactions in January 2026, with 89.2% of services priced between $0.01 and $0.10, a micropayment range that's impossible for Visa or Mastercard.
As AI adoption accelerates, and it is accelerating, despite the market panic, demand for digital-native payment layers grows proportionally. The more AI agents operate, the more they need rails that blockchain provides: credible neutrality, transparent execution, programmable money, verifiable ownership.
These developments advance based on operational efficiency, not speculative positioning. Tokenization projects don't care whether Bitcoin is at $60,000 or $120,000. Stablecoin adoption doesn't pause because Ethereum drew down 30%. On-chain settlement systems get built because they're better than the alternative, not because prices are going up.
The structural advantages remain relevant independent of leverage cycles and sentiment swings.
Watching What Comes Next
Several variables will clarify whether last week was the bottom or just the beginning.
Software multiples will tell us whether the adjustment is complete. If valuations stabilize around 6-7× revenue, markets have finished repricing SaaS for an AI world. Further compression means the pain continues.
Private credit commentary matters, particularly from BDC earnings calls. Watch for covenant stress, non-accrual rates, PIK usage trends. This is where extension risk becomes credit risk, where liquidity problems become solvency problems.
Crypto volatility patterns signal whether leverage has cleared. The $84,000 level for Bitcoin represents institutional ETF cost basis, sustained trading below this maintains sell-the-rally bias. Above it, psychology shifts.
Correlation dynamics between Bitcoin and equity indices will show whether crypto reasserts idiosyncratic behavior or continues trading as high-beta tech. Declining correlation toward the 0.2-0.3 range would signal the diversification thesis recovering. Sustained correlation above 0.7 means crypto remains just another leveraged tech bet.
AI capital expenditure guidance from hyperscalers deserves attention. Any meaningful shift in tone around return timelines or capacity utilization carries implications for semiconductors, cloud infrastructure, and the broader technology investment thesis.
Market structure always reasserts itself once leverage clears. Volatility always fades. These are constants.
The variables are different this time. We're not just watching a leverage cycle unwind. We're watching markets try to price a world where AI changes the fundamental economics of how businesses operate, how people work, and what assets are worth.
Last week, everything became one trade. The question now is whether markets are repricing temporarily for a leverage flush or permanently for a world that's already different from the one we priced six months ago.
The answer will determine whether you look back at last week as a generational buying opportunity or the early warning of something much larger unwinding.
Drop me a line:
What do you think? Do you like this? Do you not like this? I would love to hear your thoughts, so please reach me at akram@span.blog
Enjoyed this? Subscribe for the weekly deep dive, or drop me a line at akram@span.blog.