The new wave of artificial intelligence (AI) startups is beginning to look very different from the one that defined the past two years.
Instead of racing to build larger, large-scale models, a growing group of companies is focusing on the systems required to make AI usable inside real organizations.
This week’s funding announcements show how quickly the market is shifting toward infrastructure that helps enterprises run AI across everyday workflows. The startups raising capital are building everything from agent infrastructure and compute platforms to governance software and industry-specific operating systems designed for actual work environments.
At the center is the recognition that deploying AI in enterprises is significantly harder. Companies need orchestration layers for AI agents, governance systems to monitor model behavior, compute infrastructure for large-scale inference and vertical software that embeds AI across industries. Investors are now backing startups that deliver these operational essentials.
Infrastructure to Run AI Systems
One cluster of new funding is focused on the infrastructure needed to run AI within enterprises.
Agentic AI startup Lyzr on Monday (March 9) raised new funding at a $250 million valuation to support companies building AI agents that interact with enterprise data and applications. The company provides tools that help developers deploy and manage AI agents securely across internal systems, an increasingly important capability as organizations begin experimenting with autonomous workflows rather than simple chat interfaces.
Compute infrastructure provider Nscale said Monday it had also raised $2 billion in a Series C round to expand its data center and GPU capacity. The company focuses on providing large-scale compute environments optimized for AI workloads, helping enterprises and AI developers access the processing power required for training and inference. As demand for AI infrastructure continues to surge, companies like Nscale are emerging as a new generation of cloud platforms designed for AI workloads rather than general-purpose computing.
Meanwhile, Nominal, a startup building software platforms for testing complex hardware systems, said on Thursday (March 5) that it had raised $80 million at a $1 billion valuation. The company’s technology helps engineers monitor, test and analyze hardware performance in industries such as aerospace, defense and advanced manufacturing.
Security and governance are also becoming central parts of enterprise AI infrastructure. JetStream Security, founded by veterans of cybersecurity companies CrowdStrike and SentinelOne, raised $34 million in February to address what investors describe as a widening governance gap as companies adopt AI. The startup is developing tools that help organizations monitor AI systems, enforce security controls and ensure models behave within defined policies.
Vertical AI Software for Real Workflows
Alongside infrastructure startups, another group of companies is building AI-native software designed for specific professional workflows.
One example is DeepIP, which raised $40 million on March 2 to expand its platform to help patent attorneys use AI throughout the intellectual property process. The company’s system assists with drafting patent applications, analyzing prior art and managing intellectual property portfolios. Legal professionals often spend significant time reviewing technical documents and regulatory filings, making intellectual property work a natural candidate for AI-assisted automation.
Another startup, Humand, raised $66 million last month to expand what it describes as an operating system for deskless workers. The company’s platform uses AI to connect frontline employees with workflows, internal communications and HR systems through mobile-first interfaces. Deskless workers represent a massive segment of the global workforce across industries such as retail, manufacturing and logistics, yet they have historically had limited access to digital enterprise tools.
These types of vertical AI platforms represent a different approach from general-purpose AI assistants. Instead of asking workers to adapt to AI tools, these companies are embedding AI directly into the software systems.
For all PYMNTS AI and digital transformation coverage, subscribe to the daily AI and Digital Transformation Newsletters.
