Research

There’s no single architectural stack for agents, as they have multiple complex components. We see three distinct stacks for agentic AI: a model training stack, an inference stack, and an agentic runtime stack.

As enterprise AI adoption accelerates, MCP provides the identity, permissioning, and policy controls needed to make AI workflows secure, compliant, and scalable—turning models into enterprise-ready agents.

We’re proud to unveil the 7th annual Enterprise Tech 30, our data-driven list of the 60 most promising private companies in enterprise technology.

Let's look at five key themes for 2025 in AI and technology: rapid reasoning advancements through test-time compute, emergence of practical multimodal AI applications, AI agents functioning as autonomous workers, increased industry consolidation, and a more open IPO window.

Last week, Databricks raised a massive $10 billion Series J, during which they also disclosed some high level financials of their business. This, coupled with Snowflake being public, gives us the opportunity to compare the financials of these two, as I’m sure every growth fund that looked at the Databricks round did.

Stripe's data shows AI startups reaching $1M revenue 4 months faster than SaaS, and scaling to $30M five times faster. Yet AI faces higher compute costs out of the gate. Does AI still have the advantage?

243 new startups pitched their companies to an invitation-only group of venture capitalists last week at the 39th Y Combinator Demo Day. Core themes for YC S24 included AI agents, voice AI agents, AI copoilots, and AI devtools.

Product-led companies, with an intense focus on creating better products that put customers at the center of acquisition and retention, thrive with a culture where every team is all-in and in constant alignment.