Written by: Jimmy Desai, CFA, MBA, Associate Portfolio Manager, International Equities, NBF
The panic gripping Wall Street isn’t about robots taking over. It’s about a far more mundane fear: is AI making traditional software companies obsolete?
Over the past week, software stocks plunged, taking the iShares Expanded Tech-Software ETF (a good proxy for the software sector) down by nearly $1 trillion in seven days.
The trigger? The unveiling of a new legal tool from Anthropic’s Claude AI model that can handle tasks across legal, sales, marketing, and data analysis spooked investors with its advanced capabilities and caused a landslide of panic selling. And it’s not just software – AI-obsolescence fears hang over other knowledge-based job roles, too. Do you need to hire an ad agency if AI cannot only draft a script for an ad, but then generate the entire video too? Do you need management consultants to pull insights from data and generate a recommendation when AI can do a significant bulk of the work for a fraction of the cost? Cue a selloff in ad agencies, consulting firms, and outsourcing firms!
These are the questions that have markets seemingly convinced that software is dead. But take a step back from the hysteria, and a more nuanced story emerges.
What’s actually happening
Anthropic’s Cowork tool underscores the push by large language models into the “application layer” and the lucrative enterprise software business as a whole. This push is enabled by AI’s powerful ability to “think” coupled with writing functional software code autonomously. Previously, usage was limited to software engineers using AI to speed up traditional software development. Now, AI has cut out the middleman: users tell AI what they want, and the model delivers. Understandably, this has made investors nervous about which software businesses will survive.
Traditionally, software companies would employ anywhere from dozens to thousands of developers to write code. Once built, the capital outlay was virtually over: they could now sell that code to thousands of customers with minimal additional costs. Need email marketing software? That’ll be $50 per employee per month, please. Want a customer relationship management database? Add another $100 per person. Our firm pays between $15,000 and $30,000 per user annually for financial information software from FactSet and Bloomberg alone.
And once users have gotten accustomed to certain software, and all their data is captive to that software’s environment, the switching costs rise. This makes customers incredibly sticky, and coupled with a per-user annual subscription model, generates highly predictable and profitable revenues. Now AI is threatening that formula.
Plus, digging deeper, AI poses an even more uncomfortable question to the software industry’s pricing model: if AI makes work more efficient, do companies still need as many workers? If AI can empower 70 employees to do the work of 100, that’s 30 fewer software licenses to buy. Sure, software companies can try raising prices, offering ever-evolving functionality, or charging based on usage rather than headcount, but there’s no guarantee that customers will accept that.
Not all software is created equal
While the market has sold software stocks off indiscriminately, our thinking is that there are different layers of risk and defensibility for the incumbents.
Compare a simple note-taking app with the software that runs your company’s payroll, inventory, and compliance system. One, you could switch away overnight. The other is deeply woven into every part of how your business operates – connected to your bank, your suppliers, customers, and housing everything from your tax filings to employee records, supply chain data, and more.
Software that is deeply ingrained in day-to-day operations and tied to many other systems has structural advantages, even in an AI world, precisely because it is already there – making switching to another system complicated, risky, and expensive. Convincing a large multinational organization to swap out their SAP system would be extremely difficult. Transitioning an organization away from Microsoft Office, near impossible.
Meanwhile, what kind of software is at risk, in our opinion? Simple products or tools that mostly help people with routine tasks like searching, summarizing, or basic analysis. AI will likely disrupt these products quickly.
What about the cloud giants?
One intriguing wrinkle in this story is that this AI shift might be beneficial for companies like Amazon, Microsoft, and Google who provide the computing power that everyone else runs on.
Traditional software is built once and reused millions of times. But AI works differently – it generates fresh responses for every request. If AI-developed tools and analysis start replacing traditional software, cloud providers might end up processing significantly more work, not less. Instead of build once, the workflow would paradoxically switch to build each time!
While this view might seem exaggerated, even if directionally correct, it would dramatically increase demand for computational power. Therefore, any loss of revenue from software companies – the traditional clients of cloud computing services – going bust might be more than offset by increased token consumption by AI models.
The catch? Cloud providers are currently spending staggering sums to build AI infrastructure. Major tech firms are planning hundreds of billions in AI-related investment. Investors are nervous about whether the spike in spending will deliver a commensurate return and hence selling these names off right now as well.
What are we doing?
As always, our investment process and philosophy require us to seek the truth beyond the headlines. While our portfolios have more capital allocated to other areas of the AI ecosystem – particularly semiconductors and semiconductor manufacturing – we continue to see selective, longer‑term opportunities in software and are maintaining exposure accordingly, even as broader market sentiment has turned more cautious (Figure 1).
Figure 1: Semiconductor vs tech-software index performance
Source: FactSet
Take companies like SAP, a long-standing holding in our international portfolio, that are strong incumbents supplying a massively interconnected ecosystem of software that informs virtually every facet of decision-making in the C-suites of global juggernauts – we believe AI will be a tailwind, not a headwind, to such players.
Similarly, in our U.S. portfolio, we believe that companies like Microsoft and Alphabet will be net beneficiaries of AI and help shape how the businesses of tomorrow operate in an AI-powered world rather than be forgotten by them.
Seeing through the noise
Not everyone is panicking. Nvidia’s CEO Jensen Huang himself called fears that AI would replace software “illogical”. And in many ways, the anxiety remains hypothetical: leading software makers like ServiceNow and Salesforce, for example, haven’t told Wall Street that AI is causing them to actually lose any customers. But is it just a matter of “when”?
The truth is probably somewhere in the middle. Software isn’t disappearing, but the industry is being forced to prove its value in a new way. Some companies will adapt and thrive. Others that built businesses around routine tasks that AI can easily replicate may struggle.
This means that the software selloff is less about imminent catastrophe and more about investors recalibrating what software businesses are worth. It’s the market’s way of saying: prove your software is essential, not just convenient.
For now, that’s created both risk and opportunity — depending on which side of the bet you’re on.
National Bank Financial – Wealth Management (NBFWM) is a division of National Bank Financial Inc. (NBF), as well as a trademark owned by National Bank of Canada (NBC) that is used under license by NBF. NBF is a member of the Canadian Investment Regulatory Organization (CIRO) and the Canadian Investor Protection Fund (CIPF), and is a wholly-owned subsidiary of NBC, a public company listed on the Toronto Stock Exchange (TSX: NA).
The opinions expressed do not necessarily reflect those of NBF. The particulars contained herein were obtained from sources we believe to be reliable, but are not guaranteed by us and may be incomplete. The opinions expressed consider a number of factors including our analysis and interpretation of these particulars, such as historical data, and are not to be construed as a solicitation or offer to buy or sell the securities mentioned herein. Unit values and returns will fluctuate and past performance is not necessarily indicative of future performance.
The securities or sectors mentioned herein are not suitable for all types of investors. Please consult your Wealth Advisor to verify whether the securities or sectors suit your investor’s profile as well as to obtain complete information, including the main risk factors, regarding those securities or sectors.