Two trillion dollars. That's how much market cap has been erased from the S&P 500 Software & Services index since its October peak. The triggers to this collapse have been Anthropic's rapidly advancing AI tools, OpenClaw AI agents and the growing earnings concerns they have raised. Anthropic just raised a massive $30B round and OpenClaw's developer got Aqui-hired by OpenAI ensuring the open-source tool will remain supported by a foundation after going massively viral in a short time.
Why it matters: As of February 2026, this collapse is forcing companies to answer a critical question: in an era of AI agents replacing traditional software, what actually makes a company defensible?
Half of it vanished in the last two weeks alone. JP Morgan called it the largest non-recessionary 12-month drawdown in over 30 years.
Jefferies coined it the "SaaSpocalypse." Goldman Sachs compared software's future to the newspaper industry, warning this may be "the end of the beginning" of a long decline. And that was before the contagion spread.
As we reported in last Friday's Tech Buzz editorial, the selloff jumped the fence from software into wealth management, private credit, insurance, real estate services, and investment banking.
Raymond James plunged 8.8%. Charles Schwab sank 7.4%. Blue Owl Capital hit an 11-session losing streak, down 26%, because its tech-focused BDC is loaded with software loans. CBRE and Jones Lang LaSalle are each down ~12%, Cushman & Wakefield down nearly 14%, Willis Towers Watson shedding 15% for the week, and even trucking stocks getting hammered after an AI logistics firm said it boosted customer freight volumes 300-400% with no increase in headcount.
As Barclays equity strategist Emmanuel Cau told Reuters: investors are in "sell first, think later" mode, "asking 'who is next' and showing no mercy for anything remotely seen as an AI loser."
Something has shifted. The onslaught of "AI Slop" and one-shotted software becoming cheap and easy is hitting the confidence of the economy. This is fast transitioning from a software story to an everything story. It just feels like a pivotal moment. Former Presidential candidate Andrew Yang claimed in a recent viral tweet that we are just 1-2 years away from mass white collar unemployment.
The AI catalysts: Anthropic's Claude Code and OpenClaw drive the disruption
The match that lit the fire: Anthropic closed a $30 billion Series G on February 12 at a $380 billion post-money valuation, one of the largest private funding rounds in tech history. Their run-rate revenue hit $14 billion, growing 10x annually for three consecutive years. Claude Code alone is doing $2.5 billion in run-rate revenue, more than doubling since January 1. This AI agent framework is redefining how developers write code. Four percent of all GitHub public commits worldwide are now authored by Claude Code, double from a month ago. Anthropic's Super Bowl spots also helped push Claude from No. 41 to No. 7 on the U.S. App Store, driving 148,000 downloads, up 32%.
The same week, CEO Dario Amodei told Dwarkesh Patel that Anthropic is "near the end of the exponential"—not a slowdown, but the endgame. A "country of geniuses in a data center" by late 2026 or 2027. Some Anthropic engineers already write zero code themselves. The company added "another few billion to revenue in January alone."
And then there's OpenClaw, the vibe-coded open-source AI agent which hit 180,000 GitHub stars in record time and launched an agentic economy overnight. This AI agent framework demonstrates how quickly code can be generated with minimal human input.
OpenClaw founder joins OpenAI's agentic team
On Sunday, Sam Altman announced OpenClaw developer Peter Steinberger has joined OpenAI after both Meta and OpenAI made acquisition offers last week.
Steinberger vibe-coded the prototype in one hour. He told Lex Fridman that agents like his will replace 80% of apps because "every app is just a very slow API now." One person. One hour. One AI agent framework. Two of the biggest companies in the world fighting over it. That's the software singularity in a single anecdote.
"Something big is happening… something much, much bigger than Covid"
If you haven't read Matt Shumer's viral LinkedIn essay yet—6,100+ likes, 800+ comments—you should. Shumer, CEO of OthersideAI and a TBPN guest, compared this moment to February 2020: the phase where a few people noticed a virus spreading overseas, but most of us hadn't grasped what was coming.
His testimony is the ground-level version of the macro story. He describes telling AI what to build, walking away for four hours, and coming back to the finished product—not a rough draft, the completed application, self-tested, self-iterated. On the February 5 model releases (GPT-5.3 Codex from OpenAI, Claude Opus 4.6 from Anthropic), he wrote that the models now display something that feels like judgment—an intuitive sense of the right call, not just the technically correct one.
His prediction: nothing that can be done on a computer is safe in the medium term. AI is coming for law, finance, medicine, accounting, consulting, writing, design, analysis. Not in ten years. The people building these systems say one to five years. Understanding the SaaSpocalypse and how AI agents are disrupting software companies is central to this forecast.
The debate in his comments was fierce. Skeptics pointed to cash burn, trust deficits, prompt injection security risks, and the gap between a demo and a maintained enterprise product. An AI professor called it "faith-based hype." A CTO argued AI is only "as good as the average human" and CEOs will adopt it not because it's better but because they'll deem it "good enough."
But the most important counterargument—and the one Tech Buzz takes seriously—is about adoption speed. Roughly 60% of American white-collar workers are at large enterprises with 500+ employees. These organizations move slowly. Status quo bias is real. Most corporate boards prefer predictable stability over volatile potential. Even if the labs ship AGI tomorrow, the vast majority of businesses would rather take their time.
This matters enormously for how we think about the Unsloppable thesis. The market is pricing in long-term disruption right now, hence $2 trillion evaporating. But the actual operational disruption will unfold over years, not months. That gap between perception and reality is where the opportunity lives.
How to become unsloppable: A new framework for 2026
Which brings us to why you're reading a different kind of Tech Buzz feature today.
This weekend on TBPN, the daily live tech show that's become required viewing for anyone in tech and finance, hosts John Coogan and Jordi Hays introduced a framework we think captures this moment better than anything else: "Become Unsloppable." Their thesis cuts to the heart of how AI agents are disrupting software companies.
Their thesis is simple and sharp. In an era where more code could be written in the next 12 months than in all of human history, every company has to answer two questions:
- Do you have a durable moat in the AI agents era?
- Are you a true beneficiary of AI?
If the only thing standing between you and a competitor was that they'd have to spend a billion dollars hiring thousands of engineers to replicate your software, that barrier is now collapsing toward zero. So what is your actual source of strength?
TBPN mapped this against Peter Thiel's four sources of monopoly power from Zero to One: proprietary technology, network effects, economies of scale, and brand. The critical update: proprietary technology by itself is no longer sufficient. If your "proprietary tech" is a particular Python script, that's going away. But network effects aren't. Platform liquidity isn't. Regulatory moats aren't.
Coogan's best example: you can vibe-code a pickup app that looks exactly like Uber. It has the map, the button, accepts payment. But if there's no one on the other side of that network to pick you up, your clone is dead in the water. You can copy the software. You can't copy the network.
What makes companies unsloppable
Companies are unsloppable if they have moats that AI agents can't touch. These include hardware like NVIDIA, AMD, and Broadcom; social networks like YouTube, Instagram, and Roblox; marketplaces with liquidity such as Uber, Airbnb, and DoorDash; IP holders like Disney and Netflix; platforms including Spotify and YouTube; data moats like Nielsen, Bloomberg, and Palantir; and cybersecurity firms such as CrowdStrike and Palo Alto Networks (JP Morgan's top picks from the carnage).
The companies that are sloppable? The ones where the entire moat was a big bag of code and a seat-based pricing model. Where the CEO has to answer "the question" on an earnings call and the stock sells off no matter what they say.
One of TBPN's sharpest observations cuts deeply. For years, there was a huge incentive for companies whose moat was not software to present themselves as software companies. Think of companies that were really marketplaces, liquidity providers, and network effects businesses. They hired the best engineers, showcased open-source projects, and talked about the cool tech. Now the reckoning arrives: "We're going to find out who can do nothing and win."
