Meet The Startup Building India’s First Photonic Quantum Computer

Computing is about transforming raw data into intelligent inputs and breaking down complex problems into simple forms. Basically, changing the impossible into reality. With innovations in technology accelerating at the speed of thought, this reality gets redefined every day. 

A new era is unfolding with computers that are superior to supercomputers and designed to move beyond the bit system of zero and one, which is fundamental to classical computing, into the infinite world of qubits, which are basics of quantum computing.   

The technology harnesses the unique qualities of quantum mechanics to solve problems that even the most powerful classical computers can’t crack. As quantum computing promises to make problem-solving and computation smarter and faster by an unimaginable multitude, investors have rushed to park their money in quantum technology. A McKinsey report estimated global public investment in the sector at $42 Bn in 2023 from more than 30 governments. 

The rise of quantum technology has started mobilising startups, too. Bengaluru-based Quanfluence chose to ride this wave of advanced technology that is taking the world by storm. The startup is the brainchild of Sujoy Chakravarty, Ravi Mehta, Biman Chattopadhya, Anil Prabhakar, Aditi Vaidya, and Sandeep Goyal.

Chakravarty is a semiconductor engineer, with most part of his career spent in chip designing since he started working for the US-based Texas Instruments, where he met Mehta and Chattopadhya.

How Quanfluence Found Its Roots

After nearly 12 years at Texas Instruments, Chakravarty moved on to build Silabtech, focussed on semiconductor IP, along with Mehta and Chattopadhya. They ran the Bengaluru-based startup from 2012 until they exited in 2018, selling it to Synopsys, a leader in electronic design automation software.

After selling it to Synopsys, he worked with the company for a few years. In 2021, the cofounder trio began exploring opportunities and decided to team up once again. For their new venture in quantum technology, they needed expertise in physics and quantum mechanics. To fill this gap, two professors joined the team – Prabhakar from IIT Madras and Goyal from IISER Mohali. They also needed someone to manage product marketing, sales, and overall business operations. This is where Vaidya chipped in as the last member of the founding team.

“We realised that the amount of computable data was growing exponentially, yet many complex problems remained unsolvable with traditional computing. One way to tackle this was through quantum computing and quantum-inspired computing,” Chakravarty said, highlighting how they ventured into this space.

During their research, they noticed that while many companies were focussed on building software for quantum computing, that software would be ineffective without the necessary hardware to support it. Given their background in semiconductor design, they saw an opportunity in hardware development. Recognising that it was too early to focus solely on software, they decided to leverage their expertise and contribute to the hardware side of the quantum ecosystem.

Quantum technology is based on the study of subatomic particles, which is known as quantum mechanics, that the system uses to calculate outputs.

The results help in cybersecurity and cryptography to safeguard sensitive government and financial data. It also plays a significant role in intelligence gathering that secures the nation and its people from external threats, and in defence applications to optimise battlefield logistics, develop autonomous systems, and in missile guidance.

An Ising machine is a specialised quantum computer designed to efficiently solve complex optimisation issues, leveraging the principles of quantum mechanics to rapidly find the best solution to a problem by exploring multiple possibilities simultaneously, where traditional computers struggle. Various Ising machines have been developed to date, including optics-based, quantum designs.

How Quanfluence’s Ising Machine Can Tackle Real-World Challenges

One of the key innovations that Quanfluence pursued was the Optical Ising Machine. Prabhakar had originally been exploring the concept in his research, and the founding team saw its potential and decided to bring it to production.

The concept of Ising machines is relatively new, with only the second international conference on Ising machines scheduled to take place in six months. This emerging approach to computation specialises in solving a specific class of optimisation problems, rather than functioning as a general-purpose computer, Chakravarty explained.

The Ising machine is designed to address problems that can be modelled using variables and their dependencies. In scenarios where there are multiple variables with complex interdependencies and an optimal solution is required, the Ising machine provides an efficient computational method. This capability has broad applications across industries that deal with large-scale optimisation problems, making it a promising innovation in computational technology.

The startup is also building quantum computers from the ground up, specifically photonic-based ones. “Our machines are 100% developed in India. While we may source general-purpose chips, fibre spools, or diodes from abroad, the core technology, chip designs, and system architecture are entirely developed here. Unlike other quantum computing approaches where components can simply be assembled, this is a full-scale technological development effort,” Chakravarty said.

The Quanfluence machine is a hardware optimiser, and it handles up to 20,000 variables, with a next-generation version in development that will scale up to 100,000 variables. To put it very simply, it solves complex optimisation problems efficiently.

Take a company like Swiggy, for example. As soon as a customer places an order, the system needs to determine who should pick it up and deliver it. The challenge is to optimise this process while keeping the customer happy by ensuring fast delivery.

Now, consider the sheer number of variables involved: multiple restaurants, scores of delivery agents, several customers, and many orders – all dynamically changing. Traditional software can attempt to solve this by breaking down the problem into mathematical equations, but this falls into a class of problems known as NP-hard. This means that as the number of variables increases, the complexity of the problem grows exponentially.

“If solving a problem with one variable takes one computing core, adding another variable requires doubling the resources to two cores, then four, then eight, and so on. Very quickly, we run out of computing power. This is why companies started building hardware optimisers that can transform the problem from an exponential growth challenge to a more manageable quadratic one,” he said.

While food delivery is one industry where hardware optimisers can make a difference, their applications go far beyond logistics and transportation. Delivery optimisation is a simple example, but this type of problem-solving extends to various industries. For instance, AI training and inference involve parametric optimisation, which can also be efficiently handled by Quanfluence’s machine.

“In finance, our machine can help detect fraud and anomalies more efficiently than traditional methods. Any field that requires optimising large amounts of data can benefit from this technology,” the founder said.

“The way it works is straightforward: you feed a problem into the machine, and it provides a solution. However, software is needed to translate problems into a format the machine understands and then interpret the output into a human-readable format. While we don’t develop this software ourselves, we can assist clients in formulating and translating their problems effectively.”

Quanfluence’s Next Quantum Leap

For potential clients, the startup is now focussing on the logistics space, which is a massive industry with clear optimisation needs. It is also exploring use cases in finance, but not limiting the scope to just these domains. The startup also has seen interest from people working on disaster management and defence applications.

The machine is available online through APIs, and developers can access it from the developer hub in the company’s website. There’s an SDK that supports Python and other languages, making it easy to integrate. Right now, Quanfluence is not charging for access because it wants to monitor the usage and scale adoption, but it will introduce a structured pricing model soon.

Back in December, Quanfluence raised $2 Mn in a seed funding round led by pi Ventures, and it is using the funds in two major areas. First, the startup is actively developing a full-stack quantum computing system. While this is a long-term project, significant advancements are expected over the next year, though the full system may take 4–5 years to become commercially available.

Second, by the end of the year, the startup plans to launch a more powerful version of its machine, capable of handling 100,000 variables. As adoption grows, it will continue scaling by bringing more machines online and offering on-premise deployment for customers requiring dedicated access.

“Within the next quarter, we expect to see commercial use cases emerging, and in the next six months, we hope to see our technology widely deployed in production systems,” Chakravarty said.

Building an advanced technology comes with significant challenges. The engineering involved is highly complex, particularly in optimising hardware and memory access. These aspects require intricate solutions and pose considerable difficulties.

Beyond that, another major challenge is efficiently transferring large volumes of data in and out of the machine. As data scales, network speeds could become a limiting factor, creating potential bottlenecks. Overcoming these hurdles will be crucial as development progresses, Chakravarty noted.

The startup will face ecosystem challenges as well. Enterprises often hesitate to integrate new, unproven solutions due to legacy systems and high switching costs. Limited awareness and technical expertise further slow down the adoption rate.

The high cost of infrastructure and a relatively small domestic market for deep-tech solutions too create barriers. Now, we will have to wait to see whether the Indian market is ready for a quantum leap.

[Edited By Kumar Chatterjee]