I have been cancelling expensive SaaS subscriptions for the last six months.
Some we outgrew. Some were one-time needs that turned into recurring charges nobody questioned. Some we just stopped opening.
But here is the thing that got my attention. More and more of what we cancelled, we replaced with AI. A reporting suite that drew charts from numbers. A design tool that applied fixed rules to images. A dashboard we were paying hundreds a month to look at. We asked an AI agent to do the same work. It did. In minutes. For pennies.
We did not make a strategic decision to leave the SaaS economy. We just kept finding we did not need it anymore. That is a very different thing, and it is a much bigger problem for the software industry.
I have been building software since 1980. I have shipped products to 350 million users. I led the teams that created FarmVille, Command & Conquer Generals, and Lord of the Rings at EA. I am not an analyst. I am not a consultant. I do not have a portfolio to protect or an alliance to maintain. I have no political agenda and no marketing agenda.
I have watched the software business evolve from shrink-wrapped cassette tapes to mainframe and workstation software, from product sales on the Mac and PC to game sales on console and PC, from web software and mobile free-to-play to subscription models. This piece is about the disruption of the software business. All of it.
What I see on the ground is real. Not theoretical. Not projected. Real and welcome.
The gatekeepers are falling.
The Gates Have Fallen Before
Every major technology shift in modern history follows the same pattern. A gate exists. Someone controls it. The gate falls. Gatekeepers panic. People on the other side get free. And the world is better for it.
Gutenberg's printing press broke the Church's monopoly on the written word. The telephone made the telegraph operator obsolete. Radio bypassed the newspaper's control over information. Television redrew the boundaries of how culture was transmitted. Desktop publishing killed the print shop. The internet demolished the publisher as gatekeeper of distribution. Online sales destroyed brick-and-mortar retail. Mobile put a computer in every pocket and destroyed the assumption that you needed to be sitting at a desk to participate in the digital economy.
I lived one of these moments personally. I built desktop publishing software. Before DTP, you went to the print shop. They had their tools, their specialized knowledge, their proprietary processes. Desktop publishing came along and anyone could design their own materials. The gatekeepers said quality would suffer, that amateurs would flood the market. Some of that was true. It did not matter. The gate was open and it was never going to close again.
The pattern is consistent across five hundred years of history. A technology appears that makes the hard thing easy. The people who controlled access to the hard thing resist. They argue quality, safety, complexity, jobs. Their arguments contain truth. It does not save them. The gate falls. The people get free. The gatekeepers either adapt or disappear.
What is happening with AI and software right now is the same pattern. Except for one difference that changes everything.
Every previous gate controlled access to one domain. The printing press opened publishing. The telephone opened communication. Desktop publishing opened design. The internet opened distribution. Each gate, when it fell, liberated one industry.
Software is the gate that controls access to building the tools that open all other gates. When the software gate falls, it does not liberate one industry. It liberates all of them. Simultaneously.
That is what is happening now. And it is happening faster than anyone expected.
The Expanding Gap
To understand why the software industry is collapsing, you need to understand how it was built.
Software did not just solve problems. It created new ones. Every piece of software you adopted required more software to manage it. You bought a CRM, which required an integration tool to connect it to your email platform, which required a reporting tool to make sense of the combined data, which required a dashboard to present the reports to your team, which required a collaboration tool so the team could discuss what the dashboard showed.
The gap between what you needed and what you could do on your own did not shrink as you adopted more tools. It widened. The more software you used, the more software you needed. The industry built itself a self-reinforcing cycle of dependency.
Think of it as a river crossing. Software was the bridge. But each bridge you built made the river wider. More bridges required more bridges.
Consider Oracle or Salesforce. You buy one piece of the platform, and then the upgrades begin. Add-ons. Extensions. Professional services. Implementation fees. Each purchase creates a new dependency, a new integration point, a new bridge to maintain. An entire industry grew up around building, maintaining, inspecting, and charging tolls on those bridges.
AI does not build a better bridge.
Think about a trip to Mars. The journey takes months. Because of that distance, you have to carry everything with you. Equipment, supplies, fuel, life support for the outbound trip, provisions for whatever you do when you land, and everything you need for the return. The preparation for the journey is itself a massive engineering challenge. Entire industries exist to support it.
Now imagine warp drive. You press a button and you are there. Press it again and you are back. The months of travel disappear. The supplies are unnecessary. The fuel calculations are irrelevant. The entire infrastructure that existed to support the length of the journey has nothing left to support.
Warp drive does not cross the distance faster. It eliminates the distance. The origin and the destination are the same point. There is nothing to bridge.
That is what I am watching happen to my own workflow. The six SaaS tools I used to connect the stages of my product pipeline each existed because the handoff between stages was friction. Design to engineering. Engineering to QA. QA to deployment. Deployment to monitoring. Each gap had a tool. Each tool had a subscription. Now agents handle the handoffs directly.
The friction is gone. The tools that existed to smooth friction have nothing left to smooth.
This is not happening at one level. It is happening at three levels simultaneously. The bridges are gone. The rivers are gone. The geography has changed. The entire landscape that made the bridges necessary has been redrawn.
We have warp drive.
The Architecture of Collapse
SaaS is not a collection of product categories. It is layers. And AI is peeling them off from the outside in.
Every tool that takes data in one form and outputs it in another. ETL tools like Zapier, MuleSoft, Fivetran. Reporting and BI platforms like Tableau, Looker, Power BI. This layer is dissolving right now. It is already gone for people building with AI.
Slack. Notion. Monday.com. Asana. These exist because humans need to coordinate with other humans. If agents do the work, who is collaborating? The tool survives. The per-seat revenue model does not.
Monitoring dashboards, alerting rules, cost optimization tools, admin consoles. All translation in disguise — taking system data and formatting it for human consumption. When agents do the managing, the interface is optional.
The actual databases. Compliance audit trails. Proprietary data networks. These survive because they are the thing itself. Salesforce is not sticky because of its software. It is sticky because of the data inside it. The data stays. The $300-a-seat wrapper does not have to.
Security, compliance, regulatory enforcement. AI is better at pattern matching, anomaly detection, and rule enforcement. What survives is the regulatory position, the certifications, the legal standing. That is a business moat, not a software moat.
The bedrock. Raw storage is physics. S3, Azure Blob, bare metal. This does not get disrupted. But the markup on storage within SaaS is enormous. Storage becomes like electricity. Essential, ubiquitous, cheap.
The progression is relentless. Translation first. Coordination shrinks. Management becomes optional. Records get repriced. Compliance survives as a business position, not a software product. Storage drops to commodity. Each layer peels off, and the one beneath is exposed as having less value than people assumed.
Every collapsing industry makes the same arguments about why it will survive. Quality will suffer. Safety will be compromised. The complexity is too deep. Think of the jobs. Regulation will protect us. Standards must be maintained. The systemic risk is too great.
These arguments always contain partial truth. They never save the industry. The partial truth becomes the basis for a rearguard action that slows the collapse but does not stop it. The quality argument buys a few years. The regulation argument buys a few more.
But the gate is open, and gates do not close.
The Numbers
The market has already done the math.
since Jan 28, 2026
trading session
software stocks in 2026
The S&P 500 Software & Services Index is down nearly 20% from its recent highs. The sector entered bear market territory in late January 2026. The trigger was specific.
Anthropic launched Claude Cowork on January 13, an AI agent that handles legal research, CRM operations, and analytics. The exact functions that SaaS companies sell as core products. A single product launch from a single company triggered a trillion-dollar repricing event. That is how fragile the valuation was.
Hedge funds have shorted $24 billion in software stocks so far in 2026, and they are increasing their positions. Short selling software has become what traders are calling the new AI trade. Dip buyers are absent from the market, which is unusual for selloffs of this magnitude.
Key Stock Declines — February 2026
Gartner: Down 71% in 52 weeks; fell 31% in a single day. An AI technical author published "Gartner is dying," arguing agentic AI automates their highest-margin advisory services.
Thomson Reuters: Fell 15.8% in one day — a record. LegalZoom: Down 19.7%. RELX (LexisNexis): Dropped 16% in a week.
Klaviyo: Down 50%+ over the past year. HubSpot: Lost two-thirds of its value. Adobe: Multiple analyst downgrades. ServiceNow: Fallen 50% from 52-week high.
The selloff spread to Asia. TCS fell 7%. Infosys fell 7.3%. Wipro dropped 3.7%. The consulting companies that implement and maintain SaaS products are downstream of a river that is drying up.
The core vulnerability: being paid by clients for information those same clients provided in the first place.
The executives push back. Jensen Huang calls it the most illogical thing in the world that software will be replaced by AI. The CEO of Arm calls it micro-hysteria. Wedbush Securities says the Armageddon scenario is far from reality. Their argument is that enterprises will not overhaul trillions of data points in existing infrastructure.
They are right about the data. They are wrong about the wrapper.
Salesforce added 6,000 enterprise customers in a single quarter, a 48% increase. 18,500 total Agentforce customers running more than 3 billion automated workflows per month. The operational numbers are strong. The stock is still down 25%. The market is not punishing execution. It is questioning whether the entire category has a future.
For those who want the full data, there is a market analysis appendix to this report. An AI produced it. In minutes. With sourced citations from financial databases, analyst reports, and regulatory filings. That should tell you everything you need to know about what is happening to the market analysis industry.
The Human Story
Enough about markets and companies and CEOs. Let's talk about people.
There is a single parent running a side business who was paying $200 a month for tools she barely understood. Design software, email marketing, a CRM, a scheduling tool, an invoicing platform. Each one solved one problem and created three more. The subscriptions are gone now. She describes what she needs and it gets built. Her business is $200 a month more profitable and ten hours a month more efficient. Nobody wrote an analyst report about her. Nobody will.
There is a retired teacher who had an idea for an educational tool, a simple interactive program that helps children learn fractions through a game. She was told by a development firm that it would cost $50,000 and take six months. On Tuesday, she described it to an AI and it existed by Thursday. She did not need a developer, a project manager, a QA tester, or a six-month timeline. She needed an idea and the ability to describe it.
Okechukwu Nwaozor is seventeen years old and lives in Nigeria. He taught himself to fine-tune AI models, gathered and cleaned his own African-focused datasets, and built OkeyMeta, an AI platform that eight thousand developers now use. He is fresh out of secondary school. The gate is open.
Cogbill Construction is a ten-person metal fabrication shop in Vidor, Texas, population ten thousand. They serve the petrochemical industry. Jeff West is a welder at Cogbill. He used Claude to build their own ERP. It tracks orders, organizes quality control documents, links every fabricated piece to its material test reports, and eliminates the manual data entry that used to flow from paper to spreadsheet to QuickBooks.
A welder built enterprise software with an AI assistant. Published in The Fabricator, a trade magazine, not a tech blog.
These stories are becoming ordinary. And that is the point.
History is full of moments like this, moments where a gate fell and millions of ordinary people walked through.
When Deng Xiaoping opened China's economy in 1978, the iron rice bowl shattered. Guaranteed state employment disappeared. But what followed was Township and Village Enterprises, millions of ordinary people in rural China starting local businesses. China's economic miracle was not built by the Party or by state corporations. It was built by ordinary people who, for the first time, were allowed to build.
When India dismantled the License Raj in 1991, millions of small entrepreneurs emerged from nowhere. The IT outsourcing industry that built Infosys and Tata was born from that liberalization. Now, ironically, those same companies are the gatekeepers being disrupted. The circle completes.
When M-Pesa launched in Kenya in 2007, it was not about Safaricom's stock price. It was about a woman selling vegetables at a market stall who could suddenly receive payment and save money from a $20 phone. She skipped the entire banking infrastructure. The gate was not opened. It was bypassed entirely.
In every case the pattern holds. The suffering is loud and the benefit is quiet. The gatekeepers have platforms and press coverage. The newly free do not. The voices with the biggest megaphones tend to be the ones with something to protect. The people walking through the open gate are too busy building to argue about whether it should have opened.
Eight Perspectives on the Same Revolution
This is not one story. It is eight stories happening simultaneously, and they feel completely different depending on where you are standing.
They are fighting for their lives in a quiet revolution. Their product was the bridge, the translation layer, the thing that sat between a human need and a human inability to meet that need directly. The bridge is dissolving. Some will try to pivot to AI-native products. Most are bolting AI features onto a product whose core function is disappearing. That is like adding a GPS to a horse-drawn carriage.
For decades, product builders were held hostage by software engineers and software companies. Engineers were the priesthood. They had specialized knowledge, proprietary languages, years of training. Now someone who has never written a line of code can build real products that matter. The exclusivity is gone. That is a threat to them and freedom for everyone else. Their fear is understandable. It is also not a reason to keep the gate closed.
The people who paid the monthly tithes. Twenty dollars, fifty dollars, two hundred dollars a month for tools that helped with basic tasks. They were the congregation of the Church of Software, paying for access to what should have been simple. These people are the first to feel the liberation. The convenience tax is over.
People who use software throughout their lives and might not even realize it. For them, the shift is invisible. Apps work better. Things that used to take five steps take one. Costs go down. The software layer between them and their goal gets thinner until it disappears. They will not write articles about this revolution. They will just live in its benefits.
The painter who needs Photoshop. The musician who needs a digital audio workstation. The filmmaker who needs editing software. The entrepreneur who needs a prototype. AI does not give them better tools. It removes the tool barrier entirely. Creation becomes direct. Intention to output, without the software middleman.
This may be the most consequential perspective. Billions of people worldwide could not afford SaaS subscriptions, could not afford to hire software engineers, could not afford the CS degree. They were locked out by four gates simultaneously: cost, education, geography, and language. AI removes all four gates at once. A farmer in Kenya who needs inventory tracking software describes what he needs and an AI builds it. Custom. Free. Immediate. In his language. For his actual problem. The developing world may skip the entire SaaS era.
A child growing up right now does not know what a subscription is. Does not know what it felt like to be told "you need a developer for that." Does not know the gate existed. For them, there is no revolution. There is just reality. They describe what they want and it gets built. That is how it has always worked, as far as they know. This is how you know the change is permanent. When the next generation does not even understand what you were fighting about.
The VP of Engineering who has been doing this for 25 years and genuinely does not believe it will happen this fast. The analyst who calls it micro-hysteria. The CTO who says enterprises will never overhaul trillions of data points. These are not stupid people. Their entire frame of reference, built over decades of real experience, tells them this cannot happen the way people say it will. The question is not whether they are smart. It is whether they are looking down. The terrain is changing while they are still arguing about the old map.
The Meta-Gate
Here is what most people are missing about this moment.
Every previous technology revolution opened one gate. Printing opened publishing. The telephone opened communication. The internet opened distribution. Each one was transformative within its domain. The world changed, but the change had boundaries.
Software is different. Software is not one gate. Software is the gate that controls access to building the tools that open all other gates.
Healthcare has been bottlenecked by the cost and complexity of building custom software for patient management, diagnostics, records, and research tools. Education has been bottlenecked by the inability to build personalized learning tools at scale. Agriculture has been bottlenecked by expensive precision farming software. Legal services, financial modeling, manufacturing process control, government operations.
Every industry that needed custom software and could not afford it or could not build it has been sitting behind the same gate.
That gate just fell.
It did not fall for one industry. It fell for all of them. At the same time.
This is not a technology story. This is not a market story. This is not even a business story. This is a civilizational shift in who gets to build and what they get to build with.
The terrain is changing while we are opening the map to read it. And that means there really is no map.
The Road Ahead
If this piece has done its job, you are now convinced that something real is happening. The evidence is in the markets, in the history, in the mechanics, in the human stories, and in the lived experience of anyone who is building things right now.
The gatekeepers are falling.
But conviction without honesty is a manifesto, and this is not a manifesto. So here are the questions I do not have answers to.
Open Questions
Do the platform companies become the new Church? If we trade subscription tithes to HubSpot for API fees to OpenAI and Anthropic, we have not eliminated gatekeeping. We have moved it. Open-source models — DeepSeek, Qwen, Llama, Mistral — are a counterweight. This is worth watching closely and worth worrying about.
Is the AI infrastructure buildout sustainable? Goldman Sachs projects $1.15 trillion in hyperscaler capex between 2025 and 2027. To justify that investment, AI companies need roughly $1 trillion in annual profit. The 2026 consensus estimate is $450 billion. That is a $550 billion annual gap. The revolution is real, but the infrastructure financing it may be a bubble on top of a genuine shift.
Will regulation become the last moat? The EU AI Act takes effect in phases through 2026 and 2027, with fines up to 35 million euros or 7% of worldwide turnover. If compliance costs are high enough, they become a gate in themselves.
Does the locked-out world actually get access? The constraint has shifted from "can you build it" to "do you have connectivity and AI access." That is a smaller gate, but it is still a gate.
What happens to the people in the middle? New roles are emerging — AI orchestrators, prompt architects, outcome designers. The valuable skill is shifting from technical execution to problem identification and clear communication. But the transition will be painful for millions of people. That pain is real and deserves to be taken seriously.
The honest answer to most of them is "I do not know yet." Anyone who tells you they have it figured out is selling something.
Closing
I started this piece with a simple observation. I have been cancelling SaaS subscriptions because AI makes them unnecessary. That observation opened into something larger.
The subscriptions are a symptom. The disease, if you want to call it that, is the dissolution of an entire economic layer that existed because building software was hard. Building software is no longer hard. Software is effectively free, and that changes everything.
I have been making software for 46 years. I have watched gates fall before. I built tools that helped them fall. I know what it looks like when an industry is in denial, when the arguments for survival are technically correct and strategically irrelevant, when the people with the most to lose are the loudest voices in the room.
This is that moment.
The gatekeepers are falling. The gates of software creation are breaking wide open. No longer do we have to pray to the priesthood of software creation to get what we need, or pay the monthly tithes.
A five-year-old can make software to do what they want. A farmer in Kenya can build tools that fit his actual life. A retired teacher can bring her idea into the world without asking permission from anyone.
The suffering will be loud. The freedom will be quiet. The gatekeepers will have platforms and press coverage and lobbying budgets. The newly free will just build things.
That is how it has always worked. Every single time a gate has fallen in five hundred years of history, the gatekeepers predicted catastrophe and the people got freedom. The gatekeepers were always wrong about the outcome being bad. It was only bad for them.
What I see on the ground is real. Not theoretical. Not projected. Real and welcome.
The world has changed. Build accordingly.