Summary

Anthropic's new AI model Claude Opus 4.6 is causing upheaval in the software industry. The model can coordinate AI agents and handle complex programming tasks entirely automatically – a signal to investors that traditional software licenses could soon become obsolete. The market reaction was dramatic: software companies like Salesforce and Atlassian lost 25 to 35 percent of their value. At the same time, new business models are emerging through AI-powered services and startup founding is being enabled by reduced development costs.

People

Topics

  • Artificial Intelligence and AI agents
  • Software industry and business models
  • Market risk analysis and stock market reaction
  • Startup ecosystem and entrepreneurship

Clarus Lead

Claude Opus 4.6 marks a technological turning point: the AI model coordinates autonomous agents that solve complex tasks without human intervention. This fundamentally threatens the classic software license business model. In the first week of February 2026, software companies lost billions in market value as investors recognized that companies could develop their own software using AI tools – making expensive licenses unnecessary. However, the development simultaneously opens opportunities for startups and specialized service companies.

Detailed Summary

Anthropic has evolved from a 2021 footnote founded with $124 million in capital to become the most valuable AI company. In its latest funding round, the company was valued at $380 billion. The business is flourishing: annual revenue increased tenfold from 2023 to 2025 to approximately $14 billion. Claude, Anthropic's flagship product, positions itself as an enterprise-friendly alternative to OpenAI's ChatGPT with particular strengths in programming.

The new Claude Opus 4.6 can process massive amounts of data in a single query and coordinate AI agents that work as a team to fulfill specialized subtasks – writing code, documenting, testing. This fundamentally changes how work gets done: CEO Dario Amodei reported that software engineers at Anthropic no longer write code themselves; AI tools handle it. The market reaction revealed deep uncertainty: shares of Salesforce, Atlassian, and other software vendors fell 25 to 35 percent. Several catalytic events intensified the crash – new models from OpenAI and Anthropic, reports of AI-driven cost savings, and autonomous agent systems.

For traditional software companies, this means existential threat. The classic model – licenses per user, regardless of actual usage – becomes questionable when companies can generate specialized tools themselves. However, ETH Professor Georg von Krogh warns against overreaction: complex enterprise software will not become obsolete overnight. Margins and business models are under pressure, however – a pattern historically observed with typewriters (displaced by computers), DVDs (Netflix), and Blackberry (smartphones).

New opportunities are emerging in parallel. Startups like Bewy automate property management processes through AI agents – reducing administrative overhead by 70 percent and personnel costs by 50–60 percent. Founder Luca Serratore calls this "Service as a Software" instead of "Software as a Service." With mid-six-figure investments, his team was able to enter Zug, Zurich, Aargau, and Lucerne. Venture capitalist Andreas Göldi of b2venture builds internal applications in hours instead of weeks – a dramatic productivity jump that enables new business models.

Key Statements

  • Technological disruption: Claude Opus 4.6 coordinates AI agent teams that enable autonomous problem-solving for complex programming tasks
  • Business model threat: Classic per-user software licenses lose relevance when companies can generate specialized tools themselves
  • Market volatility: Software companies lost up to 35 percent of market value in February 2026; investors see structural change
  • New opportunities: Reduced development costs enable startups to grow faster; specialized AI services replace licenses
  • Transition process: Established providers like Salesforce must integrate AI features; complete displacement occurs gradually, not immediately

Critical Questions

  1. Data Quality/Validity: How robust are Claude Opus 4.6's performance benchmarks under real enterprise conditions with legacy systems and proprietary data formats? Are there independent evaluations, or do performance metrics come exclusively from Anthropic?

  2. Conflicts of Interest: To what extent does Anthropic itself benefit from market panic and associated investments in AI security and enterprise solutions? What financial relationships exist between Anthropic and venture firms investing in AI startups?

  3. Causality/Alternatives: Can the software stock crash be attributed solely to Anthropic's model, or do overvaluations, interest rate expectations, and general tech volatility play equally weighted roles? Could established firms successfully transform their business models by integrating AI agents rather than becoming obsolete?

  4. Feasibility/Risks: What practical barriers remain – change management, legacy system integration, regulatory compliance, data protection – that could slow rapid adoption by SMEs? How realistic is Serratore's scenario of 70-percent automation in complex administrative processes?

  5. Time Horizon Uncertainty: Dario Amodei says software will become "cheap or free" – but without temporal specificity. Do scenarios differ significantly across software segments (CRM vs. ERP vs. specialized tools), and how informative is this distinction for today's investment decisions?

  6. Adoption and Lock-in Effects: How strong are switching costs and organizational path dependencies in existing Salesforce/Oracle deployments? Can startups with AI agents really compete more cost-effectively when customer retention and support ecosystems are considered?

  7. Societal Consequences: The article mentions job losses only marginally (Serratore: "still far away"). How concrete are the employment risks for software developers, administrative and support roles, and how seriously does the industry address retraining responsibility?

Bibliography

Primary Source: Die Angst um eine ganze Industrie – Neue Zürcher Zeitung, Malin Hunziker, 21.02.2026

Supplementary Source References in Article:

  • Report on OpenAI's acquisition of an Austrian AI agent system
  • WEF video interview with Dario Amodei (January 2026)
  • Wired Magazine: Anthropic founding story (garden tent scene, 2021)

Verification Status: ✓ 21.02.2026

This text was created with the support of an AI model. Editorial responsibility: clarus.news | Fact-check: 21.02.2026