Agentic AI and the quiet shift beneath market pricing
Artificial intelligence capabilities continue to advance at speed, yet corporate adoption remains cautious. As a result, a significant portion of potential productivity gains is still unrealised. This disconnect has been visible in equity markets over the past year. In 2025, the software sector underperformed the broader market, despite accelerating technological progress under the surface.
Since October last year, software-as-a-service, a core pillar of the listed software universe, has come under sustained pressure. Investor concern has centred on the disruptive implications of agentic AI and the perceived vulnerability of traditional subscription-based business models. Over this period, the S&P 500 Software and Services Index declined by more than 15 percent, while the broader S&P 500 rose by close to 5 percent. Valuations compressed sharply, even as underlying operational metrics in parts of the sector continued to improve.
At the same time, innovation has accelerated rapidly outside of market narratives. The open-source AI agent OpenClaw, previously known as Clawdbot and Moltbot, spread virally through developer communities and accumulated a large number of GitHub stars in a very short space of time. GitHub, a platform where developers collaborate on code, uses stars as a proxy for interest and adoption. The speed at which OpenClaw gained traction illustrates how quickly agentic AI capabilities are evolving beyond experimental use cases.
What distinguishes this new generation of AI agents is autonomy. These systems are no longer limited to rulebased automation. They can maintain context and objectives over time, observe changes in their environment, intervene when required and learn from experience. In practical terms, this allows them to collect and update data from multiple sources, identify and correct errors, generate reports and distribute outputs automatically.
More advanced use cases include coordinating with other systems or AI agents to check availability, make reservations, authorise payments within predefined limits and update internal records. Decision-making is guided by generative AI rather than static workflows, enabling contextual judgement, multi-system orchestration and high operational independence.
This creates a strategic tension for companies. Many are caught between the fear of being disintermediated by AI and the pressure to produce a transformative “killer application”. In reality, history suggests that success rarely comes from full-scale reinvention. It is more often driven by organisational realignment. Firms that redirect processes toward AI-enabled demand and delegate repetitive, low-differentiation tasks to intelligent systems can unlock productivity gains quickly, without waiting for artificial general intelligence or dismantling existing business models. Human capital is freed to focus on higher-value activities long before the technology reaches its theoretical end state.
Capital markets tend to misread this phase of technological change. Long-term end states such as artificial general intelligence, mass job displacement and winner-takes-all platforms are frequently priced aggressively, while the intermediate phases where tangible economic value is created are overlooked. This is not new. In the 1990s and early 2000s, technologies developed by companies such as IBM, Microsoft, Intel and major telecommunications operators enabled large-scale business process automation. This was not the final form of the internet, yet enterprise resource planning systems, workflow automation and operational optimisation delivered substantial efficiency gains. Companies like Walmart achieved structurally lower cost bases and greater scalability well before internet-native models such as Amazon ultimately reshaped retail.
The lesson is that value accrues meaningfully in these middle stages. Investors who focused only on eventual end states missed years of margin expansion and cash-flow growth as incumbents improved efficiency without disrupting their core models. A similar pattern is emerging today. Softwareas-a-service and other labour-intensive service sectors have experienced valuation compression driven by AI-related uncertainty, even as fundamentals improve for firms that integrate AI effectively into their operations.
Cybersecurity provides a clear example of current mispricing. The sector has been pulled lower alongside broader software, yet the rise of autonomous AI agents operating across multiple systems significantly increases the importance of identity management, access control and behavioural monitoring. Securing agents, permissions and decision pathways becomes mission-critical well before AI reaches any final form. This creates a strong structural case for rising demand for next-generation security solutions, with positive implications for both revenue growth and margins.
Consulting and professional services follow a similar logic. These businesses are unlikely to disappear abruptly. As early as 2026, many firms are positioned to replace a meaningful share of repetitive, labour-intensive work with AI, both internally and for clients. This shift has immediate implications for margins and free cash flow, and over time should support higher valuations rather than structural decline.
Macro and market context
The current macro environment is characterised by moderating inflation, resilient services activity and financial conditions that are restrictive enough to impose discipline, but not tight enough to force demand destruction. In the euro area, headline inflation of around 1.7 percent year on year and core inflation close to 2.2 percent indicate that disinflation has progressed materially without pushing the economy into contraction. In the United States, services activity remains firmly expansionary, with leading indicators consistently above the 50 threshold, pointing to continued demand for labour and services rather than an outright slowdown.
This combination is highly relevant for the AI transition. Periods of positive, but slower, growth have historically favoured productivity investment over aggressive hiring. Wage growth has eased from post-pandemic extremes but remains sufficiently elevated to keep pressure on operating margins, particularly in labour-intensive service sectors. In this environment, AI adoption is driven by economic necessity rather than experimentation. Firms deploy AI to stabilise margins, reduce repetitive labour input and improve operational efficiency within existing structures, rather than to pursue disruptive reinvention. From a market perspective, this macro backdrop explains both the emergence and persistence of the valuation divergence observed in software and services.
In a regime anchored by persistently positive real interest rates and disciplined liquidity conditions, markets have become structurally less willing to capitalise distant growth narratives and more focused on near-term cash generation and execution. As a result, software valuations have adjusted not because operating conditions have deteriorated, but because uncertainty around longdated AI outcomes is being discounted more aggressively. This has created a disconnect between improving internal efficiency at company level and how those gains are currently reflected in equity prices.
Our perspective
We view the current phase as a classic mispricing of the middle stage of a technological cycle. Markets are discounting long-term uncertainty around artificial general intelligence and platform disruption, while underappreciating nearterm productivity gains that are already measurable. History shows that value is created first through margin expansion and efficiency gains, long before endstate technologies are fully realised.
In our view, companies that integrate agentic AI into existing workflows, particularly in software-as-a-service, cybersecurity and professional services, are positioned to deliver tangible cash-flow improvements over the next two to three years. These gains are occurring in an environment of stable demand and moderating inflation, which supports earnings visibility rather than speculative growth. Capital, in our assessment, should be focused on businesses monetising AI today, not those priced on theoretical futures.
Simonis Storm is known for financial products and services that match individual client needs with specific financial goals. For more information, visit: www.sss.com.na
