Software eats AI: verticalization drives value

What happened

Recent Q1 2025 earnings from major tech companies confirm AI's growing importance across the market landscape, with optimization opportunities moving through the entire value chain. Microsoft reported strong overall AI adoption while maintaining a 41% operating margin despite massive AI investments. International Business Machines Corp. disclosed their AI business has reached over $6 billion in total contract value, up 15% quarterly and now representing 17% of their services revenue. Meta's AI-enhanced advertising solutions generated 19% year-over-year growth, outpacing their overall revenue growth of 16%. Amazon noted their AI business has a "multibillion-dollar annual revenue run rate" growing at triple-digit percentages year-over-year, with investments spanning both hardware optimization and specialized applications.

Impact on our Investment Case

Optimization moving across the AI value chain

The focus on optimization is evident in both hardware and software components, though with distinct investment implications for each.

On the hardware side, optimization has been ongoing for some time. Amazon's custom Trainium 2 chip delivers "30-40% better price performance" than general-purpose GPUs, while Microsoft has increased AI infrastructure performance by 30% at constant power consumption while reducing their "cost per token" by more than half. These efficiency gains significantly outpace the 15-20% typical in general computing but represent a continuation of established hardware optimization trends.

As we have already argued, the more compelling frontier now appears to be software optimization. The so-called "DeepSeek moment" underscores this shift, demonstrating that architectural innovations in software can deliver efficiency gains that far outpace those achieved through hardware scaling alone. Amazon highlighted being "the first cloud service provider to make DeepSeek R1 generally available as a fully managed model"," recognizing software optimization's growing importance.

This software-driven approach to efficiency is creating new investment opportunities with potentially superior risk-adjusted returns compared to the broader hardware category.

Vertical solutions leading software optimization

Within the software category, sector-specific vertical AI solutions (i.e., the evolution from generic AI capabilities to industry-specific solutions tailored for particular sectors) are demonstrating the most promising economic profiles.  

IBM reported margin expansion in Software (370+ basis points) and Consulting (280+ basis points) largely attributed to industry-specific AI implementations -– far exceeding the single-digit expansion in their traditional services. The adoption patterns for specialized vertical solutions indicate strong customer traction. IBM saw 40% of Fabric customers adopt real-time intelligence capabilities within 5 months – compared to typical enterprise software adoption rates of 15-20% annually. IBM's average AI engagement size has increased substantially, with consulting-led implementations representing 80% of their $5 billion AI business (up from $3 billion last year).

Microsoft's sector-specific vertical cloud solutions are growing at 25-30% compared to 17% for overall Azure, demonstrating how industry-tailored offerings can significantly outperform general cloud services.

This acceleration in sector-specific solutions demonstrates how domain expertise combined with AI capabilities is creating substantial new revenue streams and driving margin improvement.

Competitive moats through domain expertise

The competitive landscape is increasingly defined by strategic differentiation based on vertical expertise. Companies are leveraging their domain knowledge to create defensible AI offerings. IBM focuses on regulated industries, Meta builds business-objective-driven advertising solutions, and Amazon combines retail expertise with cloud infrastructure. Each approach turns industry knowledge into a competitive advantage that general AI providers struggle to replicate.

Perhaps the strongest advantage for vertical solutions is their ability to create industry-specific data network effects. IBM's healthcare solutions incorporate insights from over 9.5 million physician-patient encounters, creating a knowledge base that competitors would take years to replicate. As these specialized solutions process more industry data, they improve in ways general models cannot match, potentially creating winner-take-most dynamics within vertical markets.

The integration ecosystem surrounding vertical AI solutions creates additional defensibility. Microsoft's 1,000+ partners building specialized solutions (up 40% in six months) create a surrounding ecosystem that increases switching costs. Regulatory alignment provides yet another advantage, particularly in highly regulated sectors. IBM's compliance-focused banking solutions reduce regulatory review times by 80%, demonstrating attractive ROI in areas that general AI struggles to address. These factors represent significant barriers to entry for providers lacking deep industry expertise.

Our Takeaway

The evolution we are witnessing across the AI value chain has significant investment implications. The evidence increasingly suggests that while optimization is occurring throughout the value chain, software -– particularly vertical applications -– offers the most attractive economics going forward.

These quarterly results (and the related market performance, which has seen a marked inflection point over the past 6 months –- see the below chart) are validating our strategic choice to favor software over hardware, with particular focus on sector-specific AI applications. The shift toward software optimization doesn't eliminate the importance of hardware entirely. Rather, it suggests a more targeted approach to hardware investments, focusing specifically on players driving significant advances in optimization like custom AI accelerators and specialized silicon. However, the most compelling risk-adjusted opportunities appear to be emerging in the software layer, particularly in applications that effectively verticalize AI for specific industries.

While we maintain selective exposure to leading hardware optimization players, we have long increased our portfolio exposure to software opportunities where verticalization offers the most compelling combination of growth, margins, and competitive sustainability.

Within the software category, companies that effectively combine AI capabilities with domain expertise to address unique sector requirements represent our highest-conviction opportunities. As the AI market matures, we expect this verticalization trend to accelerate, further reinforcing our allocation approach that prioritizes software innovation with an emphasis on vertical applications.