Ed Ludlow

So we're looking to invest in the building blocks of that AI infrastructure that are increasingly valuable.

So think about applied materials,

building the equipment that builds the chips that are bending the laws of physics and almost making impossible possible.

Companies like Synopsys, which sells the software that designs the chips.

Just think about chips being 30 times more complex

to design over the next five years, going from 200 billion transistors to a trillion on a chip, more power, more performance, more workloads, more agentic.

So we're looking at those areas of the AI value stream that are durable, again, designing the chips, building the chips, and then that ecosystem, clearly companies like Amazon, Google, Microsoft have scale.

Companies Broadcom and Custom Silicon, we think they have a long runway of growth.

So we're looking for those value streams that are going to help supply the compute needed for this economy long-term to go agentic, which is more inference, which is more adoption.

Going agentic.

That has upended some trades when it comes to software.

I was lucky enough to speak with PagerDuty CEO and chair Jennifer Tejada yesterday, and she was talking about how she sees the stickiness of certain software companies such as hers.

Just take a listen to what she said, Todd.

Somebody once told me that enterprise software is a little bit like a tattoo.

PagerDuty is about digital resilience, Todd, but her argument is that their moat is the infrastructure that they've spent a lot of money building out and the trust that they have with their customers.

Where are you seeing opportunity in software?

Where are you concerned about in software?

Well, where we have some concerns is we recently exited investments in ServiceNow and Workday.

Again, good companies, just the forward range of outcomes got too wide for us.

So we have backed away from those.