Building DeepTech ventures is inherently difficult. Apart from solving challenges which are within one's control such as building tech / team / product etc, a pivotal aspect influencing the success of such ventures, often underestimated, is the element of “timing”. The strategic alignment of product launch with market readiness can be a deciding factor in the trajectory of a deeptech venture.
While the concept of electric cars is quite old, it wasn't until now that the conditions became favorable for their mass adoption with innovation in battery technology, falling chemical prices & global shift towards sustainability. The success of Tesla and the broader acceptance of EVs exemplify how timing is instrumental in steering the fate of deeptech ventures.
Just as timing is important for the success of deeptech ventures, it becomes even more critical for a fund like ours to have a strong view on this. At pi Ventures, we identify such themes through the demand supply curve. This framework helps us strike a balance between innovation and market preparedness. You can read more about this framework here.
Since our inception in 2016, pi Ventures has undergone a notable evolution in investment thesis in an effort to keep the timing on our side.
As we stand today, here are our key observations/focus areas for the next few years:
- AI - We started as an AI focussed fund at a time when building a good chatbot was immensely complex, today it’s much simpler as AI is getting more & more democratized. With rapid advancement in underlying AI technology, the value/defensibility in building AI applications has gone down. We have already talked about themes within AI which excite us: Generative AI & AI Infra
- Material Science - Our thesis around the physical innovation side of things has evolved as we see a shortening of the S curves in these domains and the confluence of market demand & technology shaping up around themes such as alternate proteins, energy storage, synthetic biology & novel materials
- Digital Deeptech beyond AI: Within the digital innovation domain, our focus has expanded beyond the realms of applied AI to AI infrastructure over the last couple of years and more recently, to encompass a more comprehensive spectrum of digital deep-tech innovation where the momentum in demand supply curve is in the right direction. We believe multiple areas within this are still in their early innings and somewhat overshadowed by the current hype cycle around AI
Some of the themes that excite us in this domain include:
i) Privacy Preserving Technologies
Data is the new oil in the 21st century. It enables extraction of insights that allow enterprises to offer personalized experiences to their customers. Enterprises often deal with sensitive customer information like personally identifiable information (PII) which needs to be dealt with securely, to safeguard it from ever increasing cyberattacks and to keep up with data privacy regulations. We have a positive outlook on the role that privacy enhancing technologies will play in securing data while enabling us to draw insights from it - we’re particularly excited by
- Solutions enabling secure ML inference (existing homomorphic encryption libraries aimed at encrypting data before sending it to an external API are slow for ML workflows)
- Solutions enabling secure data collaboration (federated learning, multi-party computation, zero knowledge proof etc.)
- Technologies such as differential privacy, fully homomorphic encryption aimed at securing PII
ii) Edge Stack
Our devices are getting more powerful which is opening up multiple applications as this enables real-time, decentralized decision-making with on-the-spot intelligence at the source of data. However building an edge native stack comes with a lot of challenges. Overcoming bottlenecks in limited computational and memory resources, power consumption, data privacy, security, and scalability is crucial to unlock the full potential of Edge AI, ensuring seamless deployment and efficient operation across diverse devices and applications.
We are looking with keen interest at solutions that will enable:
- Efficient memory utilization at edge
- Deployment of large models via compression
- Faster inference
- Edge native databases
- Dynamic edge-cloud optimization
iii) Quantum Enablers
In the landscape of technological advancements, quantum computing has emerged as a game-changer, promising unparalleled computational capabilities. Outside of the innovation at a hardware level, we think solutions would emerge that would enable a seamless transition from classical to quantum computing. Some themes which we are excited by include quantum algorithm development, quantum model simulation & quantum cryptography.
In an ever evolving tech & market landscape, we are looking to hear from enterprising founders who are building differentiated technology under the umbrella of digital deeptech both within AI as well as beyond AI. Do reach out!