1300 166 138 Tell us what you need
News / AI Agents

AI agents wipe US$850B from global software stocks

By Monika Tantau • Published 8 months ago
AI agents disrupting global software stocks market 2026

Key takeaways

• US$850 billion wiped from global software stocks in days following Anthropic's Claude Cowork plugin release
• Capgemini Research Institute estimates AI agents could generate US$450 billion in economic value by 2028
• ASX-listed Decidr developing DecidrOS for autonomous agent coordination across workflows
• FortifAI building Nol8 infrastructure for enterprise AI with proof-of-concept outcomes reported
• Decidr partnerships include Now Book It, ELMO Software, Go1 and The Growth Faculty

Most Read

The per-seat software model that underpins most enterprise technology spending is facing a new question: what happens when AI systems can complete the work previously handled by multiple employees? February’s market selloff offered an early signal, with hundreds of billions of dollars erased from software company valuations in just a few days.

The catalyst was Anthropic’s release of Cowork plugins for its Claude platform. The tools are designed to carry out real workplace tasks — including contract review, financial analysis and compliance checks. For investors, the development reinforced a growing view that agentic AI is more than a productivity upgrade. It could fundamentally reshape how organisations buy, deploy and pay for software.

Anthropic’s Claude plugins spark selloff

Anthropic’s Cowork plugins differ from standard generative AI tools in a key way. Rather than simply answering prompts, agentic systems can accept a goal, break it into subtasks, connect to external tools and complete work with minimal human supervision.

That distinction carries major implications for software vendors. Traditional SaaS platforms charge customers per user seat, but if a single AI agent can perform tasks that once required several employees, the economics of enterprise software change quickly. Investors reacted swiftly as those implications became clearer.

The market selloff erased roughly US$850 billion from global software stocks within days, according to Stockhead. The speed of the decline suggests markets are beginning to price in a scenario where agentic AI adoption accelerates faster than many software vendors can adjust their products and pricing models.

Economic value projections and adoption timeline

Research from the Capgemini Research Institute estimates that AI agents could generate around US$450 billion in economic value by 2028, although widespread enterprise deployment is still at an early stage.

The shift in value may not flow primarily to traditional software application providers. Instead, it may favour companies building AI agents, data orchestration layers and the infrastructure required to support continuous AI workloads. Systems processing large volumes of unstructured data around the clock create latency and compute requirements that many enterprise environments were not designed to manage.

For organisations reviewing their software portfolios, the transition introduces uncertainty. Seat-based licences negotiated today may not reflect how work is completed in two or three years. Companies experimenting early with agentic workflows could ultimately gain structural cost advantages as the technology matures.

SaaS software companies facing AI agent disruption in Australia

ASX players position for the shift

Australian companies are already positioning themselves for the transition to agentic AI. ASX-listed Decidr is developing DecidrOS, a platform intended to coordinate autonomous agents across business workflows. The company is testing integrations with platforms including Salesforce, HubSpot and Xero.

Decidr has also announced partnerships with hospitality booking platform Now Book It, along with ELMO Software, Go1 and The Growth Faculty. These collaborations suggest some Australian enterprises are beginning pilot programs to explore how agent-driven systems might change day-to-day operations.

FortifAI, also listed on the ASX, is focusing on the infrastructure layer. Its Nol8 technology is designed to address data and latency constraints that appear when AI systems operate continuously on unstructured enterprise data. FortifAI reports significant compute and response-time improvements in proofs of concept and plans enterprise trials ahead of a broader commercial rollout.

SaaS economics under pressure

The per-seat pricing model that dominates enterprise software may face pressure as agentic AI increases individual productivity. If a single worker using AI agents can complete tasks that once required a team, organisations may find themselves paying for software seats that are no longer necessary.

That dynamic could trigger licence renegotiations and gradually shift bargaining power toward enterprise buyers. In response, software vendors may need to explore consumption-based or outcome-based pricing models to remain competitive as customers test agentic alternatives.

Agentic AI’s infrastructure demands

The February market reaction highlighted a broader question for enterprise technology leaders: whether existing software stacks are capable of supporting always-on AI systems.

Signals to monitor include enterprise trial outcomes from companies such as Decidr and FortifAI, changes in SaaS renewal behaviour and any move toward consumption-based pricing among major vendors. Capgemini’s projection of US$450 billion in potential value by 2028 provides one reference point for assessing how quickly adoption unfolds.

For Australian organisations, partnerships between ASX-listed vendors and enterprises such as Now Book It, ELMO and Go1 may offer early insight into both the benefits and the challenges of implementation. The technology remains in its early stages, but the scale of the market reaction suggests investors expect the shift to arrive sooner than many had anticipated.

Looking to stay visible as AI reshapes search? Learn how Generative Engine Optimisation helps Australian businesses appear in AI-generated answers.

Scroll to Top