Andy Kau is a partner and managing director at venture fund Walden International. A member of the Komprise Board of Directors, we caught up with Andy to get his take on key trends in the tech startup world today.
How did you get into venture capital and how have your investing priorities changed in the last 2 to 3 years?
I have been at Walden International for the past 31 years. I started my career as an engineer and management consultant on the East Coast and moved to California when my wife started at Stanford. I interviewed with several VCs when we got here and ended up at Walden. In those days it was a relatively small industry. Over the last few years our focus has been in three areas, starting with semiconductors, which has been our bread and butter for decades. Secondly, is ML and AI and we’ve made several investments in the sector. That speaks to the impact of Komprise. And big data is the third sector: how it’s being used, how it has changed the industry and how it is tied to ML and AI.
How do you see the latest announcements by the big companies (Nvidia, Amazon, Microsoft, Open AI, HPE, IBM) affecting the startup community right now?
LLMs are still new in the big scheme of things. People are still searching for profitable use cases. Aside from code writing, which is clear cut, companies are still determining how to apply this to broader sectors like travel or financial services. As a VC we’re trying to understand this on the fly. We also do a lot on the infrastructure layer and Komprise fits in well in that space. This is analogous to our investing in the semiconductor space, whether it’s making more efficient chips, better packaging or memory architectures to support these high-powered, GPU-based data center racks that are being deployed like crazy.
If you can now use the cloud for all your AI needs, though, why wouldn’t a company do that instead of building their own infrastructure?
Companies often like to keep the important data and processing on premises. There is a huge push to do confidential computing which will allow you to shield for example, PII data. In those cases, a hybrid cloud environment will work. I think that most companies will manage AI on premises or in a hybrid environment.
Komprise can address all three use cases for AI: on-premises, hybrid and cloud. Komprise brings the ability to take data from anywhere and via a single pane, look at all the data and take actions on it. This is a super important capability.
Any predictions on the evolution of AI tech in the next 12-18 months?
The AI wave is unstoppable. People need to get used to evaluating outputs from AI and understanding the risks of using information from these different sources. That’s a given. The questions really come in per what specific applications are possible. Medicine is one of the areas where we will see profound change, such as with AI-assisted drug discovery, helping make diagnoses and figuring out optimal treatments. AI can do a huge job of assisting doctors all the way down to the individual patient level. But how do you reduce costs enough so that AI gets pervasive? That comes down to infrastructure—and a lot of that has to do with power. The data centers are sucking up power like crazy. I saw an estimate that 20% of all incremental new power usage in Europe is going into data centers. Therefore, we’ll need efficient GPUs and rack systems and high-speed networks. In some ways these are the picks and shovels.
Moving into data management, Komprise is innovating to build more bridges between unstructured data and AI. How do you see the challenges and opportunities here?
Komprise is like two pieces. First, it’s dealing with the explosion of unstructured data. Companies have data all over the place and they want to more intelligently manage, group and store their data. Komprise has the migration and tiering technology to do this very well. Komprise secondly offers a way to tag data and put it into the right workflow. The ability of Komprise to take vast amounts of unstructured data and establish metatags and put the data into places where it is useful is going to be more important.
What’s holding up AI progress in enterprises the most right now in your view: funding, governance ethical and security concerns, outdated systems infrastructure – or something else?
When you say AI, it’s a broad area from GenAI to traditional ML. In the GenAI space, corporate customers are super cautious with what they put out there. The hallucination issue is real and is one of the biggest barriers to widespread adoption. Executives don’t want to create falsehoods and then deal with a huge PR backlash. On the ML side, it’s about using all the advances in algorithms to discern patterns in the data—anything from motor reliability to sales numbers. We are seeing this area really take off.
What do you like to do in your free time?
I enjoy doing a lot of things, but they have to fit in around the edges of my work-related stuff. I have a dozen investments spread across the world, so there’s also a lot of travel. I enjoy being immersed in the outdoors. We joined the Yosemite Conservancy Council, so we try to get up to the park several times a year. We do a lot of hiking, often with our golden retriever. We recently did hikes in Tanzania, New Zealand and soon the Dolomites. I think it’s important to have a deep appreciation and relationship with nature. It grounds me.