Startups on the Edge

Deep tech startups in India seem to be fast catching up with their American peers, thanks mostly to the AI smarts and electronics design skills, they have honed over the past several years.

What sets them apart, in particular, is their ability to code what some call “tiny-AI”, shrinking sophisticated algorithms without losing their functionality, and burning them on to software-defined AI hardware that packs more power in smaller spaces and consumes lesser energy.

“Optimizing code for the edge is a critical need and we are quite adept at it and we have been doing it for some time now,” says Aftab Hussain, Assistant Professor at the Center for VLSI and Embedded Systems Technologies (CVEST), IIIT-H. Indian experts are on par with the best in the world while skills in embedded systems design, knowledge of processors and architecture, and memory control are in demand for edge computing.

Industry 4.0 demand

The demand for such skills, and clientele for edge computing in general, is robust thanks to 5G network rollouts across the world, the buzz around the Internet of Things (IoT), the Industrial IOT (IIoT), and the proliferation of sensors and connected devices. This is throwing up troves of data, in particular for Industry 4.0 applications, which need processing power untethered to the cloud for real-time data capture and analytics, either on the device or as close to it as possible – in other words, at the Edge.

Indians have been good at writing code for analyzing images and video mostly used in surveillance, healthcare diagnostics, autonomous driving, and the like. Graphics processing units (GPU)s can run AI algorithms to achieve all this. But they are power hungry while data capture and analytics devices on the edge also need a small footprint. Enter software defined hardware customized for the edge to run specialized algorithms that don't have to ping the server every time data has to be analysed.

Global consulting giant McKinsey estimates edge computing hardware presents a $175 billion to $215 billion opportunity for companies by 2025. Of this the travel, transport and logistics segment alone comprises nearly half. Industry body Nasscom says research expects the global edge computing services market to cross $28 billion by 2025 growing at a CAGR of 54%.

Not surprisingly, the market is highly competitive with almost all the biggies from Amazon Web Services (AWS), Cisco Systems Inc., Intel, GE, Huawei, to HP and Microsoft, to name just a few, in the fray for a piece of the action. Most of them also offer platform services to deep tech developers.

But AI is turning out to be a great leveler and size does not really matter where writing sophisticated machine learning code is concerned. So much so, the teams of 20-40 machine learning experts seem to be equally geared up for the game.

MORE THE MERRIER

There are some 350 odd Indian startups in this space most of them focused on Industry 4.0 solutions with use cases for almost everything from autonomous driving and drones to remote monitoring, micro-datacenter operations, healthcare monitoring and even agricultural applications using sophisticated deep learning algorithms and customized chips.

“AI has now become main stream and Indian startups are strongly placed to tap the Industrial IOT space”, says Pradeep Vajram, CEO and Executive Chairman AlphaICs, adding India’s ample engineering talent puts it in a position of advantage.

Co-founded by Vinod Dham, the father of the Pentium chip, AlphaICs, has developed what it calls the Real AI Processor (RAP), a breakthrough architecture for chips embedded with deep learning algorithms.

This proprietary technology is ripe for deployment on edge applications for autonomous cars, drones, robots etc. And interestingly, the company, with operations in Bengaluru, is competing with chips from the likes of Nvidia, which is a pioneer of sorts in embedded computing systems designed for accelerating machine learning applications.

Likewise, Hyderabad-based Manjeera Digital Systems has also developed its own proprietary processor architecture which it calls Universal Multifunction Accelerator (UMA). Incubated at the IIIT-H’s Centre for Innovation & Entrepreneurship (CIE), the company already counts the likes of ISRO, Intel, Astra Microwave Products and OmniPhy among its customers.

“We have products across the spectrum and are seeing good traction”, says Dr. Venu Kandadai, Co-Founder and CEO of Manjeera. “UMA is a generic processor which means we are not limited by market segments and as such we are targeting customers across the spectrum including healthcare, fintech and space and defense”, he adds. The processor is so versatile that it is amendable to an exhaustive range of applications ranging from multimedia (audio, video, image) processing, to data security (encryption, decryption, hashing), to graphics, algo-trading, or for that matter even general computing are possible. 

Market research and consulting firm Grand View Research says the Asia Pacific will witness the highest CAGR leading up to year 2025 in edge computing with heavy investments in IoT and cloud technologies slated for the region. Telcoms are playing a big role in this, it says. For instance Reliance Jio and Cisco Systems are collaborating to deploy edge computing to develop a mobile Content Delivery Network (CDN) on Jio’s networks to enhance and optimize video experience. AWS has added two additional edge locations in India apart from its existing four new services for IoT, according to Grand View.

VC INTEREST

Both AlphaiCs and Manjeera, are scouting for VC funds to scale up the same as several other Indian edge computing startups. Having proved added impressive IP to their repertoire, they now need to add resources and, in some cases, capacity to supply systems to customers. 

Interestingly, some of the large tech companies too have begun taking positions in the space with strategies for investments, potential M&As, partnerships, and product portfolio expansions, to strengthen their position both with skills and IP.

According to one estimate globally there have been some 60 acquisitions in the edge computing space over the past five years up to the year 2019, with data center AI and security related deals accounting for almost half of them. While most of this activity has been overseas, perhaps it is time now for Indian edge computing startups to pull some VC attention in 2020.