An exceptionally brutal day on Wall Street. Software names getting hit again today after the sell-off yesterday amid these fears of AI disrupting the industry. Software stocks are seeing one of their worst declines in recent history, but almost no one is explaining the real reason why. Which is crazy, because what’s happening right now will reshape our lives over the next few years, especially if you own any software stocks or your job involves a computer. This isn’t just another tech stock sell-off. It’s a massive shift that will make some investors very rich and crush the portfolios of everyone who chose to ignore it. So in this video, I’ll walk you through the SaaSpocalypse, the agentic AI breakthroughs that triggered it, and where to invest to get rich without getting lucky.
Table of Contents
1. Introduction
2. What is Happening
3. Agentic AI Breakthroughs
4. Companies at Risk
5. Companies that will Benefit
6. Key Takeaways
7. External References
8. Frequently Asked Questions
Your time is valuable, so let’s get right into it. First things first, it’s totally normal to feel anxious if you’re watching your stocks get hammered while the mainstream media says that software is dead. but this is exactly the kind of moment that creates huge opportunities for investors who slow down, take the time to understand what’s happening, and make moves based on facts and data while the rest of the market panics. That’s exactly what this video will help you do, and I’ll break it down into four parts. First, what actually triggered this meltdown in software stocks? Second, which companies are the most at risk? Third, how bad things could actually get for them? And finally, which stocks are set to win big as a result.
But let’s start with what’s causing software stocks to crash in the first place. On January 30th, Anthropic quietly shipped a legal plugin for Claude Cowork, which is essentially a 200-line open-source text file that tells Claude how to review contracts, analyze non-disclosure agreements, compare clauses according to a legal playbook, and draft compliance summaries. Basically, this free prompt and workflow does the kind of routine legal work that law firms usually hand to junior associates and paralegals that use giant and expensive online platforms for research like westlaw and lexus nexus within days of this clawed plugin going live software as a service stocks collectively lost almost 300 billion dollars in market cap including companies that many of us use and invest in like adobe salesforce service now hubspot and intuit which is why this sell-off is being called the sas apocalypse but But here’s what actually changed, and this is the part that almost everyone is missing.
Agentic AI Breakthroughs
After this plugin and other AI agents showed that they could chew through routine document work, KPMG, which is one of the big four global accounting firms for many of the world’s largest companies, turned around and told their own auditor, Grant Thornton UK, that if AI is making audits cheaper and faster, they shouldn’t be paying 2024 prices anymore, and if it isn’t, they’ll find a firm where it is. So KPMG explicitly used AI as leverage, and forced a 14% cut on their six-figure auditing fees overnight. This dynamic is about to repeat everywhere, because once a client can point to an AI workflow that clearly reduces time and people needed for a service, they won’t just renegotiate that new AI add-on. They’ll renegotiate the entire core contract.
- AI agents are capable of performing routine document work
- AI agents can analyze non-disclosure agreements and compare clauses
- AI agents can draft compliance summaries
- AI agents can reduce time and people needed for a service
- AI agents can renegotiate the entire core contract
And that’s where the classic pay-per-software seat and pay-per-billable-hour model really starts to break. And that’s just the beginning, because agentic AI workflows have made some massive breakthroughs in just the last few weeks. First, AI agents aren’t just fixing typos in code anymore. They’re shipping serious, production-grade software on their own. Anthropic ran an experiment where they spun up a swarm of 16 Claude Opus 4.6 AI agents, pointed them at a blank codebase, and told them to build a C compiler in Rust, which is a core piece of critical software.
Expert Opinion: The Future of AI
The future of AI is not just about automating tasks, but about augmenting human capabilities. AI agents will become an integral part of our daily lives, from assisting in routine tasks to providing strategic insights.
Alex, AI Expert
Over about two weeks, those agents wrote around a hundred thousand lines of code that can run a mainstream operating system, handle popular real world apps like databases and video tools, and passes almost all the standard stress tests that you’d expect from some serious infrastructure software, all for only around $20,000 in AI spend. This would have taken a human team around a year and cost over a million dollars once you include benefits, management overhead, and so on. The key to making this all possible is something called needle in a haystack retrieval. Opus 4.6 can scan a million tokens of text and still pull out the right snippet about 76% of the time, which is roughly three times better than the next best model.
Companies at Risk
In plain English, it can hold around 50,000 lines of code in its head and reason about how all the pieces fit together, the way a senior engineer who built the system from day one would, not like a new developer skimming through it for the first time Just one year ago getting an AI model to to code for 30 minutes without falling apart was impressive Now we have swarms of AI agents running for two weeks straight and doing work that you’d normally hire a whole team of senior systems engineers for. And once that’s possible, SaaS companies charging premium prices just to support their massive headcounts starts to look a lot less attractive.
| Company | Industry | Risk Level |
|---|---|---|
| Salesforce | CRM | High |
| HubSpot | Marketing and Sales | High |
| Monday.com | Project Management | High |
| LegalZoom | Document Generation | High |
According to MarketUS, the global artificial intelligence market is expected to almost 19x in size over the next nine years, which is a compound annual growth rate of 38.5% through 2034. But many of the companies building next-generation AI applications are not publicly traded. Think about the 90s and early 2000s. Companies like Amazon and Google went public very early in their growth cycle, but today, they’re waiting an average of 10 years or longer to go public.
Companies that will Benefit
That means investors like us can miss out on most of the returns from the next amazon, the next google, the next nvidia, that’s where fundrise comes in, the sponsor of this video.
- Nvidia: leader in AI training and inference
- AMD: alternative to Nvidia for GPUs
- Broadcom: focuses on networking chips and custom ASICs
- Samsung, SK Hynix, and Micron: benefit from explosion in demand for advanced memory
- Taiwan Semiconductor Manufacturing Company: manufactures chips for Nvidia, AMD, and Broadcom
Their venture capital product lets you invest in some of the best tech companies before they go public, venture capital is usually only for the ultra wealthy, but venture capital with fundrise gives everyday investors access to some of the top private pre-ipo companies on earth.
Key Takeaways
With an access point starting at ten dollars, they have an impressive track record already, investing almost 400 million dollars in some of the largest, most in-demand ai and data infrastructure companies, so if you want access to some of the best late stage companies before the ipo, check out venture capital with fundrise using my link below today.
- AI agents are capable of performing routine document work
- AI agents can analyze non-disclosure agreements and compare clauses
- AI agents can draft compliance summaries
- AI agents can reduce time and people needed for a service
- AI agents can renegotiate the entire core contract
All right, these agents aren’t just writing code in a vacuum, they’re starting to behave like mini managers and security teams in real companies, rakuten, the online shopping and rewards platform, plugged Claude Opus 4.6 into their engineering issue tracker, and it closed 13 tickets by itself, and reassigned another 12 to the right developers.
External References
Across a 50-person team working across 6 different code bases, all in a single day.
Just to be clear, it didn’t help close out 13 tickets. It wrote the code, tested it, and pushed it to production by itself. And when it came to assigning the other tickets, it checked the backlog, decided who should tackle what, and even knew when to escalate a decision to a human instead of guessing. Under the hood, Anthropic has a feature called Teams of Agents. One lead agent breaks projects into tasks, spins up specialist teammates, and they coordinate through a shared task board, with statuses like Pending, In Progress, and Completed, while messaging each other directly when they need help. So this is less like a single chatbot, and more like a small software company living inside everyone’s computer.
Frequently Asked Questions
What is Agentic AI?
Agentic AI refers to the use of artificial intelligence to automate tasks and make decisions on behalf of humans.
How does Agentic AI work?
Agentic AI works by using machine learning algorithms to analyze data and make decisions based on that data.
What are the benefits of Agentic AI?
The benefits of Agentic AI include increased efficiency, improved accuracy, and enhanced decision-making capabilities.
What are the risks of Agentic AI?
The risks of Agentic AI include the potential for job displacement, bias in decision-making, and cybersecurity threats.
A project manager, a few engineers, and a quality assurance tester all collaborating at machine speed. On the security side, Anthropic dropped Opus 4.6 into a sandbox that was connected to developer tools and asked it to look for problems in open source software, without telling it how to do security research. And it still surfaced over 500 previously unknown high-impact vulnerabilities. That’s the kind of work you’d normally give to expensive security consultants. For investors, the big takeaway is that AI agents aren’t just helping human knowledge workers. They’re starting to replace entire layers of mid-level coordination and analysis that today’s software-as-a-service businesses are built on. And it’s not just software companies.
Two CNBC reporters with zero coding experience sat down with Anthropics tools and built themselves a working Monday.com-style project manager, complete with boards, statuses, calendars, team assignments, and email integrations. By the end, it was even pulling in emails, surfacing missed invites and unsigned documents and acting like a personal assistant instead of a static website. Oh yeah, and they built it in under an hour for $15 in compute costs. So AI agents can work in teams to make serious infrastructure software for 2% of the cost in 2% of the time. They can manage teams of real engineers and hunt down hundreds of software issues, and they can even help non-technical people build personalized versions of popular SaaS tools in one afternoon for the price of a couple coffees.
The market isn’t just reacting to one legal plugin or one bad earnings print. It’s waking up to the idea that a huge chunk of today’s software and service revenue is built on expensive human headcounts, generic UIs, and seat-based pricing in a world that’s rapidly shifting towards agents, automation, and personalized software That what causing the software stock apocalypse And now that we have that context we can talk about which companies are most at risk And if you feel I earned it consider hitting the like button and subscribing to the channel. That really helps me out, and it lets me know to make more content like this. Thanks, now let’s talk about which companies are most at risk. Most software as a service companies are built on three big assumptions.
You bundle a bunch of commoditized features behind a slick interface, you charge per human seat, and you grow by adding more users, not by radically increasing the value per user. AI agents attack all three of these assumptions. One agent can do the work of several people, or at least centralize their work into a tool, so that customers need fewer seats. AI agents can recreate many generic workflows in-house, so companies won’t pay for every little feature from third-party vendors. And of course, businesses can keep adding more of their logic directly into those agentic workflows over time, instead of buying yet another app. That’s why software valuations are compressing. It’s not that software is suddenly useless, it’s that the market no longer believes in the pay-per-seat pricing model.
So, the companies most at risk are the ones focused on lots of basic features, connected by a nice UI that charge per seat, especially in categories like CRM, project management, marketing and sales, generic ticketing and helpdesk, and simple document generation. Think about companies like Salesforce, ServiceNow, HubSpot, Monday.com, and LegalZoom. Any system with economics that depend on lots of humans clicking through lots of workflows. And things can get pretty bad for these companies pretty fast. Broad software indexes are down around 15% over the last few weeks, and SaaS-focused funds are down by more than 20% year-to-date.
Forward price to sales multiples went from around 9x to around 6x, which are levels we haven’t seen in close to a decade, as investors price in slower growth, margin pressure, and the risk of key workflows either moving elsewhere or getting rebuilt internally using AI agents. And just to be clear, all these risks multiply together.
For example, if a company has to cut their price per seat by 20%, and they sell 20% fewer seats, and they get their forward price to sales ratio slashed from 9 to 6, their stock price just fell by 57 percent.
The companies not at serious risk are the ones where the work itself actually happens or where the files and the ecosystem actually live on their platform.
For example, Adobe’s creative stack, where Photoshop, Premiere, and Firefly sit at the center of how brands actually make their images, their video assets, and their marketing content, and where generative AI is being wired directly into those workflows instead of competing with them.
Or Figma, which isn’t just a design app, It’s a multiplayer whiteboard for entire product teams with live collaboration, shared design systems, and a massive ecosystem of plugins and integrations, making it the place where product decisions get made.
Or companies like Palantir, whose Foundry and AIP platforms act as real-time systems of record for an enterprise’s operational data and specifically are built to host and control AI agents, not be replaced by them. Said another way, these are the tools where real work happens. where the content and the data are created and edited, and where agents are actually able to interact with the business. So AI tends to make these platforms more useful, not less. But even Adobe, Figma, and Palantir still have some execution risk. They still need to ship agent-first workflows, move beyond pure per-seat pricing, and prove they can actually grow revenue per customer in an AI-first world. So here’s a simple checklist to see if a software company is at serious risk of being disrupted by AI agents.
One, if it’s priced mostly per seat. Two, if one AI agent could realistically replace several of those seats. And three, if the product is mainly a user interface or workflow tool, rather than being where the data, files, and development actually take place. Any company that checks those boxes is probably going to lose a lot of business to agentic AI. But on the flip side, any company not checking those boxes could be set to win big from these agentic AI breakthroughs. So let’s talk about those next. If you’ve been watching this channel for a while, you saw this moment coming, because I’ve been talking about agentic AI for years now. That’s why this channel focuses on semiconductors, AI infrastructure, and AI-focused software platforms that are built on top of them.
Exactly the kinds of companies that benefit big time from breakthroughs in agentic AI. Semiconductors are the chips that every serious AI agent ultimately runs on If companies start replacing seats and billable hours with agents these chips will be in even higher demand Nvidia is still the default choice for AI training and inference with over a 90 share of the data center GPU market. Chip architectures like Hopper, Blackwell, and Rubin, plus the networking and software stack that comes with them, make it very hard for enterprises to switch away from Nvidia. As agent swarms scale from pilot projects to 24-7 production, Nvidia will be one of the biggest beneficiaries of every AI workload. AMD is the main alternative to NVIDIA for GPUs.
If cloud providers want pricing power and a second source for their supply chains, AMD is where they’ll go. Broadcom, ticker symbol AVGO, focuses less on GPUs and more on everything around them. From high-speed networking chips and custom ASICs, to the specialized switches that tie racks of chips together in dense AI clusters. As agentic workloads get more distributed and bound by network speeds, Broadcom will benefit because they sell the data center switch chips that remove those bottlenecks. Another bottleneck is memory. AI agents need high bandwidth memory on GPUs and huge pools of DRAM at the blade and rack levels. Companies like Samsung, SK Hynix, and Micron directly benefit from the explosion in demand for advanced memory.
And of course, the Taiwan Semiconductor Manufacturing Company, ticker symbol TSM, is the foundry that manufactures chips for NVIDIA, AMD and Broadcom. So it doesn’t really matter which chip designer wins in this SaaS apocalypse, since TSMC will be the one making their chips anyway, as well as the chips for hyperscalers like Amazon, Microsoft and Google. As far as AI infrastructure goes, Amazon, Microsoft and Google all win for the same reason. They’re where AI agents actually run. These three companies control almost two-thirds of the world’s cloud infrastructure and each company has its own agentic stack on top. Bedrock on AWS, Azure AI and OpenAI for Microsoft, and Gemini and Vertex for Google.
So every time a company replaces a SaaS tool with an agent, that extra compute, storage, and networking spend shows up in these cloud services. Companies like CoreWeave, Nebius, and Iren are the smaller and more volatile versions of that same theme. They’re building AI-focused data centers and renting out GPU time to model builders, enterprises, and hyperscalers that need extra capacity. CoreWeave is a pure-play and NVIDIA-backed AI data center operator. Nebius is scaling dedicated AI clouds for governments and regulated industries. And Ironman is turning its massive power contracts into dense AI compute clusters. These companies are much more concentrated bets than the hyperscalers, so make sure to check out my previous videos covering their upsides and their risks.
Vertiv, ticker symbol VRT, sells the picks and shovels of the data center itself. Power systems, cooling racks, and other critical hardware that lets hyperscalers, neoclouds, and big enterprises physically set up those dense AI clusters. As racks get hotter and more power-hungry thanks to GPUs and ASICs, Vertiv plays a more central role in bringing them all online. And then there’s the AI-first software that sits on top of all this infrastructure. Like I already mentioned, Palantir’s platforms are becoming the AI operating system for complex organizations, Foundry and AIP sit on top of messy but mission-critical data and let customers build AI workflows directly into verticals like supply chains, defense, healthcare, and energy. Their U.S.
commercial revenue is already growing by triple digits because AIP is all about deploying AI on live data, not selling another random app. And cybersecurity leaders like CrowdStrike are evolving into AI-focused security systems where agents help detect, triage, and even solve threats in real time. CrowdStrike’s Charlotte AI and agentic features become more important in a world where fleets of agents are talking to sensitive systems and datasets around the clock instead of human users that stop working at 5pm. Hopefully, this video helped you understand what’s really driving the SaaS-pocalypse, beyond all the clickbait headlines, and why hyperscalers are willing to spend hundreds of billions of dollars on AI infrastructure. Agentic AI is not just another hype cycle.
It’s a structural shift that will make some investors rich and crush the portfolios of people who choose to ignore what’s happening. If you made it this far into the video, it’s pretty clear which side of that line you’re on. And if you want to see what else I’m investing in to get rich without getting lucky, check out this video next. Either way, thanks for watching and until next time, this is Ticker Symbol U. My name is Alex, reminding you that the best investment you can make is in you.
Leave a Reply