Software is the layer where AGI's economic value is realized. Hardware provides the compute; software captures the revenue. The AGI buildout creates two simultaneous mega-waves: (1) explosive demand for the infrastructure software that trains, serves, and orchestrates AI models — cloud platforms, data pipelines, MLOps, observability, and security; and (2) a restructuring of every vertical application as AI agents replace workflows that previously required human labor.
Not all software companies benefit equally. Cloud hyperscalers and AI platform providers sit at the epicenter — they sell the picks and shovels to every AI builder. Cybersecurity benefits from an expanded attack surface that scales with AI adoption. Data infrastructure companies see surging demand as the quality and volume of training data becomes a competitive moat. Meanwhile, some traditional SaaS vendors face existential risk as AI agents disintermediate their workflow-centric products. This report maps each sub-sector's AGI positioning across 58 investable US-listed names.
AI/ML platforms provide the software stack for building, training, fine-tuning, and deploying machine learning models. Foundation model providers train and serve the large frontier models (LLMs, multimodal) that underpin the entire AI ecosystem. This layer includes both the model providers themselves and the MLaaS platforms that host and orchestrate models. Revenue comes from API inference fees, cloud GPU consumption, and enterprise licensing of model access.
Hyperscale cloud providers operate the massive compute, storage, and networking infrastructure that powers AI workloads globally. AWS, Azure, and GCP collectively control ~65% of the cloud market. They sell GPU-as-a-service for training and inference, managed AI services, and the underlying IaaS/PaaS that every AI application runs on. Their capex on AI infrastructure is running at $200B+ annually across the big three, building data centers purpose-built for AI.
Data infrastructure companies provide the platforms for ingesting, storing, transforming, and querying the massive datasets that power AI. This includes cloud data warehouses/lakehouses (Snowflake, Databricks), streaming platforms (Confluent), and data integration tools. AI training requires petabytes of curated data, and every enterprise deploying AI needs clean, governed data pipelines. The "data is the new oil" thesis becomes literal when AGI can exploit any dataset it accesses.
Observability platforms collect, analyze, and visualize logs, metrics, and traces from distributed software systems. As AI applications scale to millions of inference calls, observability becomes critical for monitoring model performance, detecting drift, managing costs, and ensuring reliability. AI-specific observability (LLM monitoring, prompt analytics, token cost tracking) is an emerging sub-category layered on top of traditional APM.
Cybersecurity companies protect networks, endpoints, cloud workloads, identities, and data from threats. The AGI era massively expands the attack surface: AI agents with tool access, autonomous code execution, model poisoning, prompt injection, and AI-powered adversarial attacks all create new threat vectors. Simultaneously, AI dramatically improves defensive capabilities — anomaly detection, automated response, and threat intelligence at machine speed. Security spend is non-discretionary and grows with attack sophistication.
DevOps/MLOps platforms manage the software development lifecycle, CI/CD pipelines, container orchestration, and the specialized workflows for deploying and maintaining ML models in production. AI infrastructure tools handle experiment tracking, model registries, feature stores, and GPU cluster orchestration. As enterprises move from AI prototypes to production, MLOps becomes the critical bottleneck — most AI projects fail in deployment, not in research.
Data labeling companies provide the human-annotated training data that supervised learning models require. This includes text annotation, image/video labeling, RLHF (reinforcement learning from human feedback) for LLMs, and domain-specific data curation (medical, legal, autonomous driving). The industry employs millions of annotators globally. As models improve, labeling evolves from simple tagging to complex evaluation tasks requiring skilled human judgment.
Vector databases store and query high-dimensional embeddings — the mathematical representations that LLMs use internally. They power RAG (retrieval-augmented generation), semantic search, recommendation systems, and long-term memory for AI agents. When an AI assistant needs to recall context, find similar documents, or ground its answers in a knowledge base, it queries a vector database. This is a new data infrastructure category that barely existed before 2023.
Enterprise AI enablers help large organizations adopt and deploy AI across their operations. This includes AI-focused consulting (Accenture, IBM), enterprise AI platforms that sit between foundation models and business processes (Palantir, C3.ai), and system integrators that build custom AI solutions. These companies bridge the gap between frontier AI capabilities and the messy reality of enterprise data, compliance, and workflows.
Vertical AI SaaS companies embed AI deeply into industry-specific workflows — healthcare, legal, financial services, real estate, construction, etc. Unlike horizontal AI platforms that provide general-purpose capabilities, vertical AI companies combine domain data, regulatory knowledge, and workflow integration to deliver AI that actually works in specialized contexts. Examples include AI-powered drug discovery, legal document analysis, financial risk modeling, and clinical decision support.
Robotic Process Automation (RPA) software automates repetitive digital tasks by scripting interactions with applications — clicking buttons, filling forms, extracting data, moving files between systems. The evolution from rule-based RPA to AI-powered "intelligent automation" combines traditional bots with LLM-based decision-making, document understanding, and natural language interfaces. This transforms RPA from brittle scripts into adaptive AI agents that can handle unstructured workflows.
Identity and access management (IAM) platforms control who (and what) can access systems, data, and resources. This includes authentication (verifying identity), authorization (granting permissions), privileged access management (PAM), and identity governance. In the AI era, IAM extends to non-human identities — AI agents, service accounts, API keys — that vastly outnumber human users. Managing what AI agents are allowed to do becomes a critical security and governance function.
API management platforms provide the infrastructure for creating, publishing, securing, and monitoring APIs. Integration platforms (iPaaS) connect disparate applications and data sources. In the AI era, every foundation model is accessed via API, every AI agent calls external tools via API, and every enterprise AI deployment requires integrating LLMs with existing business systems. APIs are the nervous system connecting AI brains to the world.
Edge computing software runs workloads closer to data sources — at cell towers, retail stores, factory floors, and in vehicles — rather than in centralized cloud data centers. Edge AI specifically runs inference models on edge devices for low-latency, privacy-preserving applications. Use cases include autonomous vehicles, industrial IoT, real-time video analytics, and on-device AI assistants.
CDN (Content Delivery Network) companies operate globally distributed networks of servers that cache and deliver content (web pages, video, downloads) to users from the nearest edge location. They reduce latency, improve performance, and absorb traffic spikes and DDoS attacks. Some CDN providers have expanded into edge compute, serverless, and security, blurring the line between CDN and cloud platform.
Traditional horizontal SaaS companies provide workflow tools across industries — CRM, ERP, HR, project management, communication, document management. These products were built around the assumption that humans perform knowledge work using software as a tool. AI agents that can directly execute tasks (write emails, update CRM records, generate reports, schedule meetings) threaten to disintermediate the workflow layer. The question is whether incumbents can embed AI faster than AI-native startups can replicate their functionality.
Each sub-sector is rated on its direct exposure to AGI-driven demand over the next 2-3 years, considering: (1) how directly the sub-sector enables or benefits from AI/AGI adoption, (2) structural moats and pricing power in an AI world, (3) disruption risk from AGI automating the sub-sector's own value proposition, and (4) revenue sensitivity to AI spending growth vs. traditional IT budgets.
Company lists focus on the most directly relevant US-listed names. Some companies appear in multiple sectors (e.g., MSFT in Cloud, AI/ML Platforms, and DevOps; GOOGL in Cloud, AI/ML Platforms, and API Management). Pre-IPO companies and recently acquired firms are noted but excluded from the unique company count. Tickers are as of May 2026. This report is for research purposes and does not constitute investment advice.