AI in Software - Infrastructure vs AI in Software - Application: Where to allocate capital for 2024?

AI in Software - Infrastructure vs AI in Software - Application: Where to allocate capital for 2024?

The advent of Artificial Intelligence, particularly in its generative forms, has irrevocably reshaped the technological landscape, creating a seismic shift in how software is developed, deployed, and consumed. For sophisticated investors and capital allocators, the question is no longer *if* AI will transform industries, but *where* within the burgeoning AI software ecosystem the most strategic and defensible returns can be found. The fundamental dichotomy lies between AI's impact on software infrastructure – the foundational plumbing and operational layers – and its manifestation in software applications – the direct user-facing tools and business process solutions. In 2024, navigating this distinction is paramount, demanding a granular understanding of value creation, competitive moats, and long-term secular trends.

As ex-McKinsey consultants and financial technologists, our analysis reveals that while the headlines often gravitate towards the dazzling potential of AI applications, the bedrock of sustainable value often resides within the less glamorous, yet critically indispensable, software infrastructure layer. This layer is undergoing its own profound AI-driven transformation, becoming more intelligent, autonomous, and resilient – a prerequisite for the scalable and secure deployment of any AI application. Capital allocation in 2024 must reflect this foundational reality, balancing the high-beta potential of applications with the enduring necessity of robust, AI-powered infrastructure.

The AI Infrastructure Imperative: The Unseen Foundation of the AI Economy

Software infrastructure, historically the 'picks and shovels' of the digital gold rush, has become the intelligent nervous system of the AI era. This category encompasses everything from databases and observability platforms to security, DevOps tooling, and network delivery. The infusion of AI into these layers isn't merely an enhancement; it's a fundamental re-architecture that enables the very existence and scalability of AI applications. AI-powered infrastructure offers automation, predictive capabilities, enhanced security, and optimized performance, addressing the complex demands of processing vast datasets and managing intricate AI models.

Consider the indispensable role of modern data platforms. MongoDB, Inc. (MDB), for instance, provides a general-purpose database platform designed explicitly for modern applications, now offering integrated capabilities for AI-powered retrieval, real-time analytics, and operational data. Its flagship offering, MongoDB Atlas, serves as a crucial backend for AI applications, handling the massive, diverse, and rapidly evolving data requirements that underpin machine learning models. Investing in MongoDB is not just an investment in a database; it’s an investment in the foundational data layer that allows AI applications to ingest, process, and act on information efficiently. Its ability to manage unstructured and semi-structured data, coupled with its flexible document model, makes it uniquely suited for the dynamic data landscapes AI applications operate within.

Similarly, the complexity introduced by AI workloads necessitates a new generation of observability and performance management. Companies like Dynatrace (DT) and Datadog (DDOG) exemplify the AI-infused infrastructure play in this domain. Dynatrace, with its software intelligence platform, leverages AI to automate anomaly detection and provide actionable insights across complex cloud environments, which are increasingly dominated by microservices and AI models. Its end-to-end observability is critical for monitoring the health and performance of AI algorithms, ensuring their reliability and efficiency. Datadog, on the other hand, offers an integrated observability and security platform for cloud applications, providing real-time visibility into an entire technology stack. As AI models become integral parts of applications, monitoring their performance, resource consumption, and potential biases becomes paramount. Datadog's AI-driven insights help engineering, operations, and security teams proactively manage these sophisticated systems, preventing outages and optimizing resource allocation for costly AI inference and training.

Security and application delivery also see AI integration as a necessity. F5, Inc. (FFIV) provides multi-cloud application security and delivery solutions. In an AI-driven world, where APIs are the conduits for model interaction and data exchange, securing these endpoints becomes exponentially more critical. F5's Application Delivery and Security Platform (ADSP) combines high-performance load balancing with advanced application and API security features, often incorporating AI for threat detection and anomaly identification. This ensures that the underlying infrastructure for AI applications remains resilient, performant, and impervious to sophisticated cyber threats. The growth of AI applications directly correlates with the increased demand for secure, efficient application delivery, making F5 a key infrastructure enabler.

Data protection and cyber resilience, especially for the invaluable datasets that fuel AI, are equally critical. Commvault (CVLT) provides data protection and cyber resilience software, enabling organizations to secure, back up, and recover data across any environment. The proprietary data used to train AI models represents an immense competitive asset, and its loss or corruption can be catastrophic. Commvault's platform, often enhanced with AI for ransomware detection and accelerated recovery, offers a robust defense, safeguarding the very foundation of an enterprise's AI initiatives. This is a foundational layer that often gets overlooked but is non-negotiable for enterprise AI adoption.

Finally, the entire software development lifecycle, particularly for AI-driven projects, benefits from intelligent orchestration. GitLab Inc. (GTLB) provides an intelligent orchestration platform for DevSecOps, offering a single application to streamline planning, coding, security, and deployment. As AI models become integrated into software, the complexity of managing their lifecycle – from data preparation and model training to deployment and monitoring – increases dramatically. GitLab's platform, increasingly leveraging AI to assist developers and security teams, enables organizations to accelerate the development and secure deployment of AI-infused software, improving developer productivity and reducing time-to-market for innovative AI applications.

While Verisign (VRSN) operates at an even more fundamental layer as a global provider of internet infrastructure and domain name registry services (.com, .net), its relevance, though indirect, underscores the absolute necessity of robust, stable underlying systems for *any* internet-connected AI application. Without a functioning, secure internet, the entire AI ecosystem collapses. While not directly integrating AI into its core offering in the same way as others, Verisign represents the ultimate 'picks and shovels' play, benefiting from the sheer volume of internet activity, regardless of its specific AI content.

The AI Application Frontier: Direct Impact, Higher Volatility

The AI application layer represents the direct monetization of AI capabilities, delivering tangible value to end-users or transforming specific business processes. This includes everything from generative AI tools, predictive analytics dashboards, intelligent automation platforms, and AI-powered customer service bots. These applications are often built upon the infrastructure described above, consuming its services to function.

The allure of AI applications is undeniable: they promise exponential growth, disruptive innovation, and direct market impact. Companies in this space are often characterized by rapid product cycles, intense competition, and a high degree of reliance on underlying foundational models (whether proprietary or open-source). Success hinges on factors like user experience, domain-specific expertise, data advantage, and the ability to articulate clear value propositions to target customers.

While we haven't been provided specific examples of pure AI application companies in the database, it's crucial to understand their symbiotic relationship with infrastructure. An AI-powered virtual assistant, for example, relies on databases like MongoDB for data storage, observability platforms like Datadog or Dynatrace for performance monitoring, and secure delivery mechanisms from F5. The success of an AI application is inextricably linked to the robustness and intelligence of its underlying infrastructure.

The capital allocation decision here involves a higher risk-reward profile. Identifying winning AI applications requires deep market insight, an understanding of specific industry pain points, and a keen eye for sustainable competitive advantage. Moats can be built through proprietary data, superior model performance, network effects, or seamless integration into existing workflows. However, the rapid pace of innovation in this sector means that today's leader can quickly become tomorrow's laggard if they fail to adapt.

Synergies and Interdependencies: The AI Flywheel

It is a reductive error to view AI infrastructure and AI applications as mutually exclusive investment categories. In reality, they exist in a powerful, self-reinforcing dynamic – an 'AI Flywheel.' The proliferation of sophisticated AI applications drives an increased demand for more robust, intelligent, and scalable infrastructure. This, in turn, enables the development of even more advanced and resource-intensive AI applications, perpetuating a virtuous cycle.

For instance, as more enterprises deploy large language models or complex predictive analytics, their need for efficient data management (MDB), real-time performance monitoring (DT, DDOG), secure application delivery (FFIV), and robust data protection (CVLT) intensifies. Moreover, the demand for streamlined AI development and deployment via platforms like GitLab (GTLB) becomes critical to maintain competitive velocity. The more successful the AI application layer becomes, the greater the pressure and opportunity for the infrastructure layer to innovate and expand.

Conversely, advancements in AI infrastructure – such as AI-driven automation in observability platforms or more efficient AI-powered databases – lower the barriers to entry and accelerate the development cycles for AI application providers. This synergy suggests that a balanced portfolio, while leaning into the foundational strength of infrastructure, should not entirely neglect the potential of the application layer, especially those applications that demonstrate clear product-market fit and a defensible competitive position.

Capital Allocation Strategy for 2024: A Nuanced Approach

For 2024, our recommendation for capital allocation leans towards a strategic blend, with a foundational emphasis on companies deeply embedding AI into their core software infrastructure offerings. This approach offers a more defensible, lower-volatility pathway to participate in the broader AI growth story, while still allowing for opportunistic allocations to high-potential AI application plays.

The Prudent Core: Investing in AI-Infused Infrastructure

Allocate a significant portion of capital to infrastructure providers that are demonstrably integrating AI to enhance their product capabilities. These are the companies providing the essential tools and platforms that *all* AI applications, regardless of their specific vertical or use case, will rely upon. Their revenue streams are often more predictable, driven by sticky enterprise subscriptions and expanding usage. Look for:

• Strong Recurring Revenue: High Net Revenue Retention (NRR) and a significant portion of revenue from subscriptions.
• Clear AI Integration Roadmap: Evidence that AI is not an afterthought but a core component of their product evolution (e.g., Dynatrace's AIops, MongoDB's AI-powered retrieval).
• Broad Applicability: Solutions that cater to a wide range of industries and use cases, providing a larger Total Addressable Market (TAM).
• Deep Technical Moats: Proprietary technology, significant R&D investment, and intellectual property that is difficult to replicate.
• Customer Lock-in / Switching Costs: Platforms deeply integrated into enterprise operations, making them difficult to replace.

Companies like MongoDB (MDB), Dynatrace (DT), Datadog (DDOG), F5 (FFIV), Commvault (CVLT), and GitLab (GTLB) fit this profile exceptionally well. They are critical enablers whose growth is tied to the overall expansion of the digital economy and the specific explosion of AI workloads. Their value proposition strengthens as AI applications become more complex and pervasive, demanding more from the underlying infrastructure.

The Opportunistic Edge: Strategic Bets on AI Applications

A smaller, yet still meaningful, portion of capital can be allocated to select AI application companies. Here, the due diligence must be even more rigorous, focusing on:

• Differentiated Value Proposition: Does the application solve a critical pain point in a unique or significantly better way than alternatives?
• Strong Product-Market Fit: Evidence of rapid adoption, high user engagement, and clear customer satisfaction.
• Defensible Moats: Proprietary data, superior algorithms, strong brand, or significant network effects that deter competitors.
• Clear Path to Profitability: While growth is key, a sustainable business model with a reasonable path to profitability is crucial.
• Management Team & Execution: A visionary and capable management team with a proven track record.

These investments carry higher risk but also promise potentially exponential returns. It requires a more active management approach and a willingness to embrace volatility. The focus should be on application providers that are not merely 'AI wrappers' but deeply leverage AI to create novel value, transforming specific workflows or industries.

"“In the AI gold rush, smart money isn't just chasing the prospectors; it's investing heavily in the engineers building the autonomous machinery and the robust railroads that make the gold accessible to everyone. The picks and shovels are now intelligent, self-optimizing platforms – and they are the bedrock of scalable AI value.”"

Conclusion: The Intelligent Portfolio for the AI Era

The distinction between AI in Software - Infrastructure and AI in Software - Application is not merely semantic; it represents a critical divergence in risk, reward, and long-term value creation for capital allocators in 2024. While the allure of groundbreaking AI applications is strong, the prudent strategy involves anchoring a significant portion of one's portfolio in the companies that are building and fortifying the intelligent infrastructure upon which the entire AI economy rests.

The companies detailed from the Golden Door database – MongoDB, Dynatrace, Datadog, F5, Commvault, GitLab, and even Verisign – exemplify this foundational strength. They are not just participating in the AI revolution; they are actively enabling it by making the underlying software infrastructure more intelligent, secure, and performant. Their embeddedness within enterprise IT, combined with the increasing demands of AI workloads, positions them as essential partners in every organization's AI journey.

In 2024, the winning capital allocation strategy will be characterized by a balanced, yet strategically weighted, approach. Prioritize the foundational, AI-infused infrastructure layer for its stability, predictability, and broad applicability. Supplement this with highly selective, well-vetted investments in the application layer, focusing on companies with clear competitive advantages and a demonstrable path to sustainable profitability. The future of AI is not just about what it can do, but how robustly, securely, and intelligently it can be built and operated. The smart money understands that the invisible strength of infrastructure is where the most profound and enduring value will be created.