Lightspeed’s Aluwi: Compute isn’t AI’s bottleneck in SEA

While concerns over AI compute shortages continue to dominate the global conversation, Kevin Aluwi, venture partner at VC firm Lightspeed and co-founder of ride-hailing giant Gojek, says Southeast Asia’s biggest bottleneck isn’t accessing chips – it’s scaling up the adoption of AI products.

In a conversation with Tech in Asia, Aluwi discusses how AI is reshaping company-building, startup investing, and the future of the region’s tech ecosystem. While Lightspeed backs startups from several sectors, he says that the VC pays attention to how teams use AI in their business.

He declined to comment on reports that Lightspeed is raising US$300 million to US$350 million for its fifth India-focused fund.

Kevin Aluwi, Gojek co-founder and venture partner at Lightspeed / Photo credit: Lightspeed

Despite his transition into venture capital, Aluwi says he still misses building companies and thinks about returning to the founder’s seat “all the time.”

This interview has been edited for clarity and brevity.

If you were starting Gojek from scratch today, what would an AI-native Gojek look like?

There are three different buckets that I would think about.

When it comes to hiring, we would probably have a much leaner team. I was in a lot of discussions back then with people who – for very valid reasons – always requested more headcount.

For a long time, the company was growing incredibly fast, and so we always needed more capacity to do more things. Asking whether or not something can be done with AI versus hiring would probably be one big change in terms of organizing and building teams.

There is also something to be said around architecting the company to be machine-readable. We had a distinct advantage relative to other companies because we had distributed teams. We had a large engineering center in Bengaluru, along with engineering centers in Singapore, and we also had operational teams everywhere.

The more immediate challenge is driving the adoption of AI-powered products and services at scale.

We had aggressive policies around documentation because not everyone was in the same office. So a lot of things were written and recorded, with lots of artifacts everywhere. I would have loved the opportunity to train AI on how the company works based on these artifacts.

Based on that, I would love to explore how executive decisions can be automated, because you have this system that can ingest data and make sense of the whole company. The whole company has already been recorded in collections of Slack messages, Google Docs, etc. I don’t know exactly what that would produce, but it would have been something really, really interesting in terms of building a strong operational cadence that is assisted by AI.