HSBC just threw cold water on the 'SaaSpocalypse' panic gripping software investors. The bank's analysts are flipping Marc Andreessen's famous thesis on its head, arguing that established software giants - not AI-native startups like OpenAI or Anthropic - will capture most of the value as artificial intelligence goes mainstream. The contrarian call comes as enterprise software stocks recover from brutal February selloffs triggered by fears that AI would cannibalize traditional SaaS business models.
HSBC is making a bold contrarian bet that could reshape how investors think about the AI revolution. While much of Wall Street has been wringing its hands over whether OpenAI and Anthropic will disrupt the entire software industry, the bank's analysts are arguing the exact opposite - that legacy software giants are positioned to win big as AI transitions from buzzword to business tool.
The timing couldn't be more provocative. Software stocks just endured one of their worst months in recent memory, with investors fleeing amid fears of a 'SaaSpocalypse' where AI agents would replace traditional SaaS applications entirely. Salesforce and Oracle both saw double-digit drops as analysts questioned whether their subscription-based models could survive in an AI-first world.
But HSBC's research team sees a different picture emerging. Their thesis centers on a simple insight: enterprise customers don't want to rip out their existing software infrastructure and start over with AI-native tools. They want their current vendors to make their existing systems smarter. That structural advantage, the bank argues, gives established players an unbeatable moat.
Oracle tops HSBC's conviction list, with analysts pointing to the database giant's aggressive push into AI-powered cloud infrastructure. The company has been quietly embedding machine learning capabilities across its entire product suite, from autonomous databases that tune themselves to AI-driven financial planning tools. More importantly, Oracle controls the data layer where most enterprise AI models will ultimately need to operate.
