At the recent HumanX Global AI Conference held in San Francisco, analysts from global investment leader UBS observed that OpenAI and Anthropic are transforming from AI model giants into significant consumers of enterprise IT budgets. According to the UBS team, the introduction of more capable large AI models by leading companies like Anthropic and OpenAI poses a substantial threat to the growth prospects of many traditional software firms. These advanced AI companies now possess the strength to capture a larger share of enterprise clients' spending.
Enterprise wallet share refers to the portion of a client's total IT, software, and digital budget allocated to a specific technology provider. UBS analysts note that budgets previously directed toward traditional software vendors are increasingly being diverted to frontier AI companies such as OpenAI and Anthropic. Whereas enterprises once allocated substantial funds to SaaS applications, BI tools, customer service software, low-code platforms, and collaboration software, many are now directly using AI application platforms like Claude or ChatGPT to build generative AI solutions and AI agents. This shift is causing a large-scale transfer of spending from traditional software providers to large AI model developers.
Market concerns regarding software stocks are not unfounded. The HumanX AI conference in San Francisco attracted around 6,000 investors, entrepreneurs, and technology executives. The event highlighted the deepening relationship between AI technology and various industries, while also sparking intense discussions about AI's impact on employment.
A key trend noted at HumanX is the significant increase in corporate investment in AI agents, extending well beyond core use cases such as Microsoft’s Copilot licenses and AI programming aids for large-scale software engineering. UBS analysts Karl Keirstead and Dean Marriott wrote in an investor report that recent enterprise surveys support the pessimistic outlook already reflected in stock prices, as well as more cautious views on SaaS and application software companies.
The urgent corporate need for efficiency gains and operational cost reduction is accelerating the adoption of two key AI application categories: generative AI apps and AI agents. Autonomous AI agents, capable of handling complex and tedious tasks, represent a major long-term trend in AI application over the next decade. Their emergence signals AI's evolution from an information aid to a highly intelligent productivity tool. According to MarketsandMarkets research, the AI agent market is projected to reach $53 billion by 2030, reflecting a compound annual growth rate of 46% starting in 2025.
The expected proliferation of autonomous AI agents like Anthropic’s Claude Cowork and OpenClaw by 2026 is no coincidence; it results from the simultaneous convergence of five key factors: model capability, tool protocols, AI developer frameworks, inference costs, and end-context capacity. At the application layer, AI agents are likely to become the dominant commercial interface, translating intelligence into action and advancing AI from answering questions to executing, collaborating, and completing highly complex multi-step tasks.
Anthropic’s recent report on agent coding trends predicts that by 2026, agentic workflows will expand from engineering teams to non-technical departments such as sales, legal, operations, and marketing. The company emphasizes that agents are not merely a feature but are becoming the default interaction paradigm for next-generation software—acting as context aggregators and action orchestrators, shifting focus from apps to intent, and from tools to full autonomous execution.
UBS learned from conference participants that OpenAI and Anthropic are increasingly viewed as best-in-class AI application software companies. The trend of clients using Claude or ChatGPT to build customized AI applications and agent-centric workflows, combined with these companies launching first-party products, heightens risks to the future profit growth of established software vendors.
Recent launches of high-efficiency AI agent products by leaders like Anthropic and OpenAI—which could replace certain functional software services at much lower cost—have contributed to heavy selling pressure on global software stocks. Since February, a pessimistic "AI disruption" narrative has taken hold, driven by concerns that viral AI agent workflows like Claude Cowork and OpenClaw could undermine the SaaS seat-based subscription revenue model. This sell-off has spread to cybersecurity, online education, traditional finance, insurance, real estate, transportation, and other sectors reliant on per-seat or labor-intensive business models—areas perceived as highly vulnerable to AI-driven disruption.
Software stocks worldwide have faced continued declines amid AI-related fears, despite increased share buybacks in the U.S. market. Investor skepticism stems from deeper concerns about long-term fundamentals and whether business models will be fundamentally reshaped by AI agents.
That said, some software companies appear better positioned to withstand the AI wave. Those deeply integrated with AI technology and focused on securing and managing enterprise data—such as Snowflake, Palantir, and Databricks—are seen as relatively safe. Analyst Keirstead noted that many surveys highlight data as a key advantage, with clients investing in modernizing and opening access to their proprietary data. This long-observed trend supports a bullish outlook on data-centric software firms. While AI risks are being scrutinized, there is little evidence from surveys that large models like Claude or GPT pose immediate threats to companies like Palantir, Databricks, or Snowflake—aside from isolated mentions of Claude replacing certain Tableau or Salesforce functions.
Regarding Microsoft, UBS reported mixed views. Some respondents expressed that Microsoft has fallen behind and that its Copilot-based AI applications have been disappointing. Others cautioned against underestimating the company, citing its strong reputation in enterprise security, access control, and governance, as well as its unique position to deeply embed AI workflows.
Finally, while Anthropic’s Claude Mythos made a splash last week, it is too early to assess its potential impact on security-focused software segments. Current UBS survey data indicates that the market has not yet reached an informed consensus on this front.