Why the software selldown poses a risk and an opportunity

Among growing tension around the ultimate winners and losers from the rollout of artificial intelligence, software companies have found themselves in investors crosshairs.

In early February, Anthropic released a series of Claude Cowork enterprise plugins for departments across a business, including finance, legal, sales, and data analytics.

The immediate market response was dramatic; software-related indices fell 6% in one session and are down over 20% year to date, with individual companies in drawdowns exceeding 50%. In Australia, the S&P/ASX All Technology Index, which has heavy exposure to software companies, is in a 40% drawdown with Xero and WiseTech each down more than 60% from peaks.

The question for investors is whether this is a reasonable market response or an overreaction representing a buying opportunity.

Volatility can reflect uncertainty rather than a structural decline

This is not the first time an AI development has triggered a sharp and rapid market selloff.

In January 2025, Chinese startup DeepSeek released R1, an open-source large language model claiming to match leading US AI performance at a fraction of the cost.

Nvidia lost $588 billion in market capitalisation in a single session, the largest single-day loss for any company in US stock market history. The Nasdaq fell 3.1%. Within 48 hours, Nvidia had rebounded 9% and the tech sector posted its sharpest two-day recovery in over two years.

Since the initial selloff this time around, the S&P North American Technology Software Index has rebounded 14.4%, although it remains 25% below previous highs.

The pattern across both episodes is consistent. A new AI capability emerges, markets de-risk broadly, and stocks are indiscriminately affected, pricing in the fear of a potential future rather than a realised one.

History broadly supports these recoveries. Following sharp Nasdaq selloffs, the index has averaged a gain of 8.7% over the subsequent three months, trading positively 80% of the time.

Where the underlying demand for a sector remains structurally intact, acute bouts of competitive uncertainty have, more often than not, proven to be noise rather than signal. However, this is not a given. Selloffs occasionally reflect something more permanent, and investors should weigh that possibility seriously.

Not all software is equal

What markets have correctly identified is that uncertainty around software earnings is higher. When you are less confident in the future impact of AI on software profits, the discount rate investors apply goes up.

Near-term earnings pressure is real. SAP fell 16% and ServiceNow dropped 11% on earnings resulted in February because enterprises are reducing seats rather than adding them, and the per-seat pricing model underpinning SaaS economics for two decades is under threat. Gartner estimates 35% of point-product software-as-a-service (SaaS) tools will be replaced by AI agents by 2030.

But the conclusion that enterprises will stop buying software does not reflect how large organisations operate. Businesses dependent on essential platforms are not going to ask a junior employee to rebuild their technology stack from scratch. The February selloff did not make that distinction. Platforms with deep data moats that are essential to workflows were impacted as if they faced the same fate as genuinely vulnerable point solutions.

Cybersecurity is a clear example. As AI tools become more powerful, they expand the attack surface rather than reduce it.

AI has introduced new threat tactics including prompt injection attacks and agent impersonation, creating new challenges for enterprises to navigate. Leading platforms like CrowdStrike and Palo Alto are well positioned here, embedding AI within their own systems to proactively identify and respond to threats in real time. Enterprises are unlikely to replace that with internally built tools or unproven alternatives. For example, Claude Code Security remains a research preview with no production track record.

Enterprise software is similarly embedded in the functioning of modern businesses. Scalable platforms managing data, workflows and compliance carry the same characteristics, proprietary datasets, deep workflow integration and switching costs that make replacement a significant operational risk. AI integration is more likely to enhance the importance of these platforms than to eliminate them.

ETF investors are responding with conviction

Trading data suggests many Australian investors interpreted the sell-off as a buying opportunity. In fact, technology-related ETFs saw a 132.5% increase in buying in February compared to January. Total buying rose to $146 million in February, up from $90 million the month prior. As a percentage of all equities ETF buying, technology allocations climbed to 6.8% in February, up from 4.2% in January (Betashares, IRESS).

The Betashares S&P/ASX Australian Technology ETF received a record $107 million in net flows in February 2026. That figure was three times higher than the previous monthly net flow record set in December 2025, and five times higher than inflows in January.

We also saw increased interest in exposures such as the Betashares Global Cybersecurity ETF, showing continued conviction in essential segments of the technology ecosystem.

The case for diversified exposure

Identifying which software companies emerge from this period as winners and which face structural decline will be genuinely difficult. Out of hundreds of companies that listed during the dot-com boom, one, Amazon, went on to substantially reward long-term investors.

This is where diversified exposure becomes relevant. Rather than concentrating risk in individual names, ETFs provide exposure to a broad basket of companies positioned across AI-driven innovation. As leading companies grow and their market capitalisation rises, their weight within an index increases naturally, investors gain greater exposure to emerging winners over time while limiting damage from those that do not adapt.

Whether the February selloff ultimately proves an entry point depends on how the structural questions around AI and software resolve.

What is clearer is that the demand for digital infrastructure, cybersecurity and critical enterprise software remains intact. For investors with a long-term view, diversified exposure to that theme, without needing to precisely identify individual winners, may be the most sensible way to navigate the uncertainty ahead.

Get stories like this in our newsletters.