As of writing on the 12th of March 2026, the iShares software ETF (IGV) is down over 27% year-to-date, and since its September 2025 peak, has fallen by more than 30% — erasing upwards of $2 trillion in market cap. For the first time in the modern era, software stocks now trades at or below the S&P 500 multiple.
The rapid pace of AI agent innovation, in particular Anthropic sparked a massive sell-off in the past couple of months and the selloff cascaded into well-known software stocks like Salesforce, Adobe, ServiceNow, Intuit, and others.
System of Records vs System of Engagement
This is perhaps where many investors got confused.
Agentic AI agents are targeting and decimating systems of engagement first. Think like this – if an AI agent can answer customer tickets, manage projects, or route communications — you don’t need Zendesk’s interface, or Asana’s board, or HubSpot’s sequences. The human-in-the-loop that justified the per-seat license disappears.
But AI agents can’t replace systems of record （or at least it will be too expensive/too little efficiencies, and it isn’t worth the effort to do so — they depend on them. An agent is only as good as the data it can access. It needs Salesforce’s 10 years of customer history. It needs Workday’s org structure. It needs SAP’s inventory data. The system of record is what gives the agent context, memory, and authority.
This is why the market selloff has been indiscriminate, but the survivability is not. Companies like Salesforce and ServiceNow sold off ~50% but they know they will have the last laugh. They’re not being replaced by AI; they’re becoming the infrastructure that AI runs on top of.
Jensen Huang said it best in his recent interview by giving an analogy about a hammer as a tool. If you need a hammer, say to do your house renovation works, would you rather use an existing tool, or would you create another hammer? Of course it makes sense to just use the hammer itself.
Two Stocks with Generational Buying Opportunities & I am Heavily Vested
There are many opportunities right now in the software stocks market.
I started buying about one to two months ago for these two companies and have a vested position of about $500,000 in total at the moment. Even though I am currently in a slight red at the moment as of this writing （down about 9-10%）， I know that not too distant in the future, these investments will print a good amount of investment return.
ServiceNow is one of the few SaaS companies where AI is already showing up in the financials, and the company has pivoted towards it, not just the narrative. It’s Now Assist agentic AI product surpassed $600 million in annual contract value, with net new ACV more than doubling year-over-year in Q4 2025, and management is targeting over $1 billion in Now Assist ACV for 2026. Meanwhile, full-year 2025 sales rose 21% to $13.3 billion, and the company ended the year with $28.2 billion in remaining performance obligations — a robust 27% year-over-year growth — giving strong visibility into future revenue. Additionally, ServiceNow’s AI Control Tower platform aims to be the central hub for developing and deploying AI agents, both from ServiceNow and third parties.
Despite the impressive results and outlook, its shares are down sharply from a 52-week high, and the companies were swept up in the SaaS Apocalypse fear — presenting what I thought is a compelling entry, as AI is an opportunity, not a threat for the company.
Constellation Software is another Berkshire alike company that has been around for many years doing vertical integration of thousands of software companies. It is more of a capital allocation play and the trust in management. Since CSU’s 2006 IPO, it has compounded at 29.3% annually, with revenue growing at a 23.5% CAGR and free cash flow at a 30.9% CAGR and I believe it will be a once in a generational buying opportunity, despite the AI risk that presents.
CSU’s model is to buy mission-critical vertical market software businesses cheaply and hold them forever. With its strong cash flow and acquisition expertise, CSU could strategically make savvy, cheaper acquisitions in a depressed valuation environment, guiding acquired firms through the AI transition. Furthermore, CSU largest expense to date in on headcount payroll and if AI is going to benefit efficiencies, we should be seeing increasing productivity and an increase in their ROA over time.
Author is vested in both positions as of writing.
