- Insights This article was written by a TIA community member. Insights pieces undergo the same rigorous editorial process that newsroom-produced articles have.
There’s more opportunity in the AI boom than chips and ChatGPT
This piece was written by Joash Lee of VNTR Capital and Rupin Ramesh of Atlas Ventures.
It’s obvious that AI is on a hot streak. What’s less apparent is how investors can pick winners within the sector.
As upstarts flood the scene, competition is intensifying, with numerous players vying for dominance across various layers of the value chain.
Not all upstarts will emerge victorious.
Where do the opportunities lie? Who will take the crown? And are we currently in a bubble? Let’s dive in.

Image credit: Timmy Loen
Segmenting the AI market
The AI value chain can be divided into four unique layers:
- Hardware: Encompasses the physical components necessary to power AI systems
- Storage: Storage solutions for the management and retrieval of data used to train AI models
- Middleware: Software tools that enable developers to fine-tune, train, and deploy AI models
- Applications: End-user solutions using existing AI models as is or fine-tuned to specific use cases

Image credit: Joash Lee and Rupin Ramesh
As you go down the value chain, you get closer to the end user. The potential for value creation also increases, which means companies can make greater margins.
This is the case for a few reasons:
- Customization at the application layer allows companies to differentiate their offerings, often justifying higher prices and increasing margins.
- Products closer to the end user directly impact user experience, so good tools can create strong brand loyalty and retain customers, leading to stable and potentially higher revenue streams.
- There is potential to integrate other offerings to increase the value proposition, allowing tools to command premium prices.
- Compared to the hardware layer, the application layer is more scalable as the cost does not increase linearly with volume, as the product is software, not hardware.
Hard going
Historically, CPUs were the main processors in data centers, carrying out tasks by handling one instruction at a time.
Storage, sweet storage
Stuck in the middle with you
Apply yourself
Are we in a bubble?
Where the winners lie
Stay ahead in Asia’s tech landscape
This is premium content. Subscribe to read the full story.
The AI boom has led to an avalanche in startups in the space. How can investors unearth the winners?
We know this is not ideal. ⌛ Sign up in 20 seconds. Cancel anytime.
Our subscriber community includes professionals from these companies:





Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.