JPMorgan reveals top ‘AI-resilient’ software to buy after crash

The Global Financial Context James Dimon

The software sector has been rattled by a sharp selloff after new AI tools promised to automate white-collar work, from tax planning to legal research. Into that turmoil, JPMorgan strategist Dubravko Lakos-Bujas has circulated a client note arguing that investors have overreacted and highlighting a group of “AI-resilient” software names to buy on the dip. His basket leans on large platforms such as Microsoft and cybersecurity specialists like CrowdStrike that JPMorgan says should benefit, not buckle, as generative and agentic AI spread.

The Software Selloff: What Sparked the Crash

The latest slide in software and data stocks did not come out of nowhere. Accountability reporting on the market reaction traces a key trigger to Altruist unveiling its Hazel AI tax-planning tool, which stoked fears that automated systems could undercut traditional fee-based services. The same coverage notes that this shock hit a JPMorgan-tracked “Magnificent 19” basket of high-growth tech and financial names, where software and finance stocks dropped in the 5 to 10 percent range as traders rapidly repriced earnings expectations.

That repricing accelerated when a separate report attributed another leg of the slide to Anthropic. According to coverage that explains Anthropic’s launch of a legal-work automation product, investors suddenly had a concrete example of AI handling tasks traditionally carried out by law firms and data providers. Yahoo Finance then documented how this wave of concern wiped roughly a reported figure of market value from the sector in about a week, with its analysis of the post-announcement selloff in software and financial stocks tying the move directly to AI disruption worries.

JPMorgan’s Take on AI Disruption

Into this backdrop, JPMorgan’s equity strategy team led by Dubravko Lakos-Bujas has tried to separate signal from noise. A client note described in detail by Yahoo Finance says the bank views the drop in the Magnificent 19 names as overdone, with Lakos-Bujas telling clients “you are over-reacting, we believe” when it comes to the impact of new AI tools on established software franchises. The note frames the selloff as part of a broader AI-driven repricing narrative, but argues that not all software revenue is equally exposed to automation.

Part of JPMorgan’s argument leans on emerging research into what some academics call agentic AI. A recent technical paper on agentic AI security describes systems that can take multi-step actions across tools and data sources, and it stresses that such systems require deterministic controls enforced through cryptography and policy across prompts, tools, data and context. JPMorgan’s analysts point to that kind of work to suggest that as AI systems become more autonomous, demand should rise for security and governance layers rather than collapse, which in their view supports select software providers instead of threatening them.

Defining ‘AI-Resilient’ Software

The JPMorgan note, as summarized in several outlets, sets out a rough checklist for what it calls “AI-resilient” software. According to one recap that highlights the bank’s preferred names, the bank favors companies with high switching costs, a core focus on cybersecurity or infrastructure, and business models where AI is more likely to enhance the product than replace it. The same reporting ties the concept directly to the Magnificent 19 framework, suggesting that even within that growth-heavy basket, some names sit on more defensible ground than others.

Finviz also cites JPMorgan projections that this resilient subset could gain roughly 15 to 20 percent over the next 12 months if sentiment normalizes after the crash. To understand why the bank sees such upside, investors have been directed to SEC filings that detail each company’s business model and risk factors. For example, filings for cybersecurity vendors describe subscription contracts, multi-year commitments and embedded agents on customer devices, all of which contribute to high switching costs that can buffer short-term volatility even as AI tools evolve.

Top Picks from JPMorgan’s Basket

Reporting on the client note indicates that Microsoft features prominently among JPMorgan’s preferred AI-resilient names. Coverage of the Magnificent 19 by a major analysis of the basket identifies Microsoft as a core holding, citing its scale in cloud computing and its integration of generative AI into products such as Azure and productivity software. That same piece points to Microsoft’s reported 200 billion dollar Azure revenue run-rate as a key data point behind the bank’s confidence that AI will deepen, not erode, its customer relationships.

CrowdStrike is another standout in the JPMorgan basket. Finviz’s recap of the note lists CrowdStrike among the AI-resilient software names, highlighting its role in endpoint security and threat intelligence. Recent 10-K filings, accessible through the SEC’s company search portal, describe how CrowdStrike deploys lightweight agents across customer devices and analyzes telemetry in a centralized cloud platform, creating a data and integration moat that would be difficult for a new AI entrant to replicate quickly. Other Magnificent 19 constituents mentioned in the note, according to the same reporting, include additional enterprise software and security vendors that share similar characteristics of sticky customers and infrastructure-level roles.

Why These Matter for Investors Now

The timing of JPMorgan’s call is central to its message. The bank’s “buy the dip” argument, as summarized by Yahoo Finance’s account of the client note, comes directly after the Hazel AI and Anthropic announcements triggered abrupt drawdowns in software and financial names. By stepping in after that shock, Lakos-Bujas is effectively telling clients that the market has priced in a worst-case scenario for many incumbents without fully considering where AI might reinforce existing platforms.

At the same time, the selloff has highlighted genuine threats to traditional software models. Reporting that details Anthropic’s legal-work automation tool notes that shares in data and education group Pearson fell as investors weighed the risk that AI systems could handle parts of legal research and training that previously required licensed content. For investors, JPMorgan’s basket is presented as a way to stay exposed to AI-related growth while avoiding the most directly threatened niches such as commoditized legal or tax-prep workflows.

Risks and Uncertainties Ahead

Even JPMorgan’s preferred names face open questions. The Finviz summary of the bank’s projections makes clear that the 15 to 20 percent upside for AI-resilient software is a forecast, not a guarantee, and it depends on sentiment stabilizing after a period of heightened volatility. Different outlets have also reported varying depths for the initial selloff, with some coverage citing drops closer to 8 percent and others pointing to 12 percent for parts of the Magnificent 19, underscoring that the exact scale of the move is still being debated. That uncertainty matters because it shapes how much “rebound” room investors assume is left.

There is also limited historical evidence on how truly resilient any software model will be in the face of agentic AI. The arxiv paper on security for agentic systems stresses that new architectures require deterministic controls across prompts, tools, data and context, yet it does not claim that specific listed companies are guaranteed winners. Similarly, SEC filings accessed through Use SEC disclosures for Microsoft, CrowdStrike and other names in the JPMorgan basket lay out risks ranging from competitive pressure to regulatory change. JPMorgan’s guidance to buy the dip, relayed through Reports on the bank’s note to investors, is therefore best seen as one informed view in an unsettled debate over how AI will reshape the software sector rather than a verdict on how the story ends.

More From The Daily Overview

*This article was researched with the help of AI, with human editors creating the final content.