Stock Markets February 24, 2026

HSBC Says Enterprise Software Will Be the Primary Channel for AI Adoption

Bank argues foundation models are ill-suited to replace core enterprise platforms; incumbents seen as best placed to commercialize AI starting in 2026

By Marcus Reed
HSBC Says Enterprise Software Will Be the Primary Channel for AI Adoption

HSBC analysts contend that traditional enterprise software will remain the primary conduit for artificial intelligence in large companies. The bank cautions that foundation models have limitations that make them unlikely to displace established enterprise platforms, and says 2026 will mark the start of meaningful monetization inside software.

Key Points

  • HSBC asserts enterprise software will remain the primary vehicle for AI adoption across the world’s largest companies, with 2026 flagged as the start of meaningful monetization.
  • The bank views foundation models as unsuitable for broad "lift-and-replacement" of enterprise platforms, arguing they may only be effective for narrow use cases such as image generation or small applications.
  • Established enterprise software vendors are seen as best positioned to embed and commercialize AI by integrating controlled, distilled intelligent agents into their platforms.

HSBC has concluded that enterprise software will continue to be the dominant route through which the largest corporations integrate artificial intelligence into their operations, even as large-scale foundation models advance rapidly.

Analyst Stephen Bersey of HSBC wrote that "software will be the primary mechanism for the diffusion of AI across the world’s largest enterprises," and identified 2026 as "the kick-off for monetization within software." The note argues that foundational AI models are "inherently flawed" for wholesale replacement of major enterprise platforms.

According to HSBC, foundation models may be serviceable for focused tasks - for example, generating images or powering small, narrowly scoped applications - but they are "not realistic for the majority of high-fidelity enterprise class platforms." The bank emphasizes that these enterprise-grade systems demand qualities and capabilities that foundation models do not naturally provide.

One specific limitation HSBC highlights is what it calls "vibe-coding," a model of development that shifts design responsibility onto the developer. The bank asserts that the leading creators of large AI models have "little to no experience creating 'enterprise class' software," meaning those firms would effectively be starting from ground zero when confronting the complexity of enterprise environments.

HSBC also notes that enterprise software has matured toward being "almost error-free with high throughput and reliability." That accumulated intellectual property - the code, architectures, and operational practices that underpin dependable enterprise systems - is not something that can be trained or replicated simply by scraping data from the public internet, the bank says.

Even in a scenario where a new entrant managed to produce comparable code, HSBC underscores the practical difficulty of displacing incumbent vendors. Those vendors operate critical systems that underpin the core operations of global companies that are accountable to shareholders, creating strong inertia against wholesale replacement.

Given these dynamics, HSBC argues established enterprise software vendors are positioned to commercialize AI first and most effectively. The bank points out that incumbents are already integrating what it calls "distilled intelligent agents" into existing platforms in controlled ways that address the shortcomings of foundation models. HSBC concludes these enterprise vendors are therefore likely to be the main beneficiaries as monetization of AI within software accelerates.


Promotional note appearing in the original release

Separately within the original material, a product promotion states that AI-driven stock selection tools claim notable returns: year-to-date, two out of three global portfolios are said to be outperforming their benchmarks with 88% in the green, and a named flagship strategy reportedly doubled its benchmark over an 18-month span, citing specific winners. This promotional content was present in the source material but is not an analytical claim about the HSBC research itself.

Risks

  • Foundation models may be effective only for narrow tasks and are considered "inherently flawed" for replacing high-fidelity enterprise platforms - impacting enterprise software and technology vendors.
  • Major AI model creators have limited experience building 'enterprise class' software, which raises execution risk for any new entrant attempting to displace incumbents - affecting cloud providers and AI startups.
  • Critical enterprise software intellectual property is not trainable on the public internet, making it difficult for outsiders to replicate incumbent capabilities and increasing barriers to entry - impacting software competition and corporate IT procurement.

More from Stock Markets

Intuit and Anthropic Forge Multi-Year Deal to Deliver Custom AI Agents for Businesses Feb 24, 2026 CSX Overhauls Data Infrastructure in Strategic Tie-Up With Infosys and Microsoft Feb 24, 2026 Keurig Dr Pepper Lifts Full-Year Outlook as Sodas Hold Strong Feb 24, 2026 Morgan Stanley Raises Booking to Overweight, Cites Durable Role for OTAs in Travel Feb 24, 2026 UBS Lowers Gerresheimer to Sell, Cuts Price Target to €12.90 Citing High Leverage and Weaker Margins Feb 24, 2026