Silicon Valley is often touted as the apex of innovation, having always received a flood of risk capital to find the next big thing. It has germinated pivotal technology companies, such as Facebook, Google, PayPal, Twitter, Salesforce, Airbnb and Uber. These companies first proved their products in local ecosystems, then scaled the US market, and eventually transcended borders.
They can claim comprehensive success in dominating the “first billion” market, and in cementing Silicon Valley as the centre of the technology world as these products and business models have been emulated by waves of startups across regions. More recently, however, the success of China-based Alibaba, Didi Chuxing and Tencent’s WeChat started to indicate unique customer behaviour in the emerging economies that comprise the next 6 billion.
These differences are becoming clearer by the day, and starting to render the copy-and-paste business model unsustainable. Both Uber and Amazon lost market share battles in China, and have hit considerable turbulence in India. Rocket Internet’s cloning model has run into far more serious trouble in India, with both online fashion house Jabong and online food retailer FoodPanda having lost significant market share.
These multinationals may well still outspend local competition, but not without innovating to create products that fit local market needs.
Given the need for cash and localisation, where is the new bastion of innovation for these next 6 billion? Our bet is Bangalore, India, given its vibrant ecosystem of talent and its proximity to the large Indian market.
India’s advantage
A talent bargain: The establishment of strategic research and development centres by US bigwigs, such as Google, Microsoft, Oracle, Cisco and now Apple, has already validated the depth of Indian software engineering talent. What is new, however, is the rate at which top-notch talent is pursuing entrepreneurship over more traditional careers. Given the living-room success stories of the past few years, social norms have changed. Studies suggest now over 10% of top Indian institutes of technology and management graduates go on to pursue startups as founders or employees. Experienced professionals, armed with deep domain knowledge, are increasingly switching to join or set up new companies. A recent LinkedIn survey listed six Indian cities among the top 10 global destinations for growth in tech talent.
Two trends are particularly material here. First, the level and quality of talent has definitely seen a big boost. The quality differential between Indian and US product engineering talent is converging. Just this last batch of Y Combinator took in three India-based teams, perhaps what has triggered its management to visit the country to set up a local chapter. Second, the cost of talent still remains significantly low compared with US benchmarks, offering a much higher potential return on investment. A fresh technology institute engineer’s salary is a quarter that of an entry level engineer in Silicon Valley.
A large market with headwinds: India has more than a billion phone-users – 20% using smartphones – and 450 million internet users, with another 200 million to be added by 2020.
The government has proved an important catalyst for the ecosystem. First, it has committed about $3bn through a fund-of-venture-funds model, which is channeled through different nodal agencies. Second, a slew of entrepreneurship friendly policies – both direct and indirect – have been instituted, including easier company setup, venture financing, tax policies, and public listings, and a more open banking licence regime to drive financial inclusion and digitise payments.
Third, and perhaps most powerfully, it has launched a public application program interface platform called the India Stack, which contains a biometrics-based identity layer, called Aadhaar, to support electronic know-your-customer, a paperless layer to support e-contracts and secure file-sharing, a payments layer to enable one-click financial transactions between any two entities, consumer or merchant, and a consent-based data layer currently being architected but which has the potential to leapfrog the most developed economies in terms of data management and governance. The Aadhaar platform has surpassed a billion users faster than even Whatsapp, which Facebook acquired in 2014 for $19bn.
AI as India’s special edge
Artificial intelligence (AI) has been in winter for decade. For a long time, even after lot of research into the area, the results were far from practicable. The way we humans learn and behave is very different from the fully or semi-supervised algorithms that were pushed up as part of AI. As a result, the outcomes were far from what could be useful in day-to-day life. That started to change in 2005, when computer scientist Geoff Hinton realised that neural nets given the availability of faster and cheaper computing power made unsupervised learning possible. Thus, deep learning was born.
Today deep learning is not only powering route optimisation, as in Google maps, but recognition of speech, objects, images and gestures. Facebook and Google algorithms convert text to meaning at the level of underlying sentiments and motivations, as in Syntexnet and Wit.ai, Facebook recognises faces, as in DeepFace, with up to 97% accuracy, almost as close as a human. As these advances demonstrate, “machines have started to learn” like the way humans’ do.
Fortunately, in the past year or so, all the major players have started to put their core AI research effort for others to use on tap. For example, TensorFlow was originally developed by researchers and engineers working on the Google Brain Team within Google’s Machine Intelligence research organisation for the purposes of conducting machine learning and deep neural networks research, but the system is general enough to be applicable in a wide variety of other domains as well. So, recently, Google opened it for everyone to use and profit from. Similarly, Facebook opened up its Bots framework powered by Wit.ai, pretty similar to a recent move by Microsoft.
The net result is a level playingfield for global startups, which is of a disproportionate advantage to Indian startups. The combination of a large and rapidly growing mobile-internet connected population along with startups being able to access their data along with open-source intelligence research means they can leverage decades of research, apply it to data, and pipe the results to build context-aware, predictive and extremely personalised business models.
This can be built on intelligent process automation and forecasting frameworks for speed and precision. Overall, these businesses will have potential for step-changes in operational efficiency and effectiveness in understanding and fulfilling customer needs.
This is especially so in India, where small screens on many smartphones can create vast pools of data to power AI. By further automating business decisions through machine learning, and surfacing them through smart, intelligent interfaces, these startups will disrupt older technologies and traditional businesses – based on heuristical approaches – and can emerge as “new category leaders”.
As a global average, 99.5% of data gathered is never analysed. Interestingly, most of this data is local and contextual, which still has deep value. Indian tech companies have access to vast amounts of data. According to a recent Cisco study, 148.9 petabytes (a petabyte is 1,000 million million bytes) of monthly mobile data traffic was generated in India in 2015 – equivalent of 410 million text messages each second. With deeper penetration of 3G and 4G LTE networks, this is expected to grow 11 times by 2020.
The critical challenge for Indian technology companies will be to harness this digital data and build “data moats” that power best-in-class predictive insights and uniquely differentiated customer experiences. To do this, they will need to engage with three levels of data:
• Big data or data generated by enterprise applications such as supply chain systems, customer relations management and order management.
• Small data or data generated by personal devices such as smartphones, smartwatches, fitness trackers revealing personal, contextual information.
• Dark Data or data that never came out before, but now is available for analysis and action, most probably due to internet-of-things sensors or devices.
Our hypothesis is that a combination of at least two of these would be required for a winning strategy that would create enough adaptive learning to outplay competitors. Capital alone will not determine who prevails, but real and defendable business value derived from machine-learnable data will. Indian technology companies must find better ways to innovate on data – either themselves or by letting other innovators access and build on their data.
Whitespace everywhere
Many sectors that crunch massive amounts of data to solve high-stakes problems, such as healthcare, financial services and insurance, security and fraud, skills training, agriculture, and commerce, will see AI action.
Retail alone is a $500bn industry projected to double by the end of this decade, and e-commerce still constitutes less than 2%. E-commerce companies have pushed adoption of mobile commerce in India – 80% is mobile, versus 50% in US. Mobile channels generate deeper and more intimate insights into shopping behaviour, with greater potential to optimise the shopping experience and influence behaviour. For example, now we can determine patterns such as shoppers with certain income and demographic profiles prefer black patterned dresses around their birthday period, but are most likely to purchase online only after visiting an offline retail outlet. Using these data patterns, consumers can be nudged with the right products at the right time. The real differentiation with bricks-and-mortar retail, which still controls an overwhelming majority of sales, is the use of data to personalise and optimise shopping experiences.
Similarly, technology can streamline supply chains by disintermediating middlemen and improving transaction economics. Customers otherwise used to painful, operationally-intensive processes can benefit from more relevance, convenience and lower costs when buying products and services. As an example, the supply chain for fresh agricultural produce has more than five intermediaries between a farmer and customer that suck up around 80% of margin. A handful of companies are working to disrupt this $70bn highly fragmented and localised market through tech-enabled operations that can not only predict demand based on day-to-day conditions, but also allocate orders to the best markets based on location-based pricing. Once catalysed, growth in any of these sectors is likely to be ferocious.
Open data vaults: window into a new innovation corridor
Leveraging data’s full potential, however, requires new models of engagement across ecosystem stakeholders. Such developments in the west have included open innovation ecosystems – pioneered by the likes of Procter & Gamble – or even the open-source movement which successfully broke through the innovation deadlocks created by Oracle or SAP to change the shape of enterprise software ecosystem, and is now threatening to displace centralised security and digital governance systems with Blockchain.
In particular, government and industry consortiums can drive the adoption of common standards in emerging areas, thereby alleviating grid-locks and expanding the market value of new solutions.
The Unified Payments Interface (UPI) project is one such very recent example, where government agencies like the Reserve Bank of India and the National Payments Corporation of India have partnered with major banks, technology providers, venture funds, and industry groups to adopt an interoperable digital payments platform that promises to unlock a wide spectrum of opportunities.
So far, however, leading Indian software product companies have been reluctant to share proprietary data. Consequently, no vibrant third-party innovation ecosystems have emerged around a data-as-a-service model.
It will be interesting to see if this changes. Significant sunk investment has already gone into obtaining data, but given its scale and scope, these players are unlikely to reap its potential alone. Instead, this presents an opportunity to create an “open data vault”, a collective pool of data and behaviour markers. Governance of such an open platform will require a clear framework for ownership, usage rights and commercials, as well as perhaps a trusted manager to implement it.
So on the consumer front, e-commerce platforms can provide mobile shopping insights, while app-based transport services can share location and movement patterns, while fin-tech and payments companies can provide financial and transaction insights, and so on. This can catapult innovation especially when additional data sources, including credit, identity and social data, are layered in. These vaults can hide personal particulars on demand, or share this information with consumer consent, and be designed to share insights, which can then be tied to actions using applied AI.
Innovation ecosystems can then be grown around these open data collectives.
We, as venture funds, can help by provide risk capital to startups working on these open data vaults, together with “unicorn” corporations and other players. In doing so, the aspiration will be to build a multilayer data ecosystem on which startups can build applied AI applications and ground-breaking customer experiences.
Big unicorns, the false dawn?
Over the past couple of years, the consumer internet sector has witnessed an explosion of aggregators and marketplaces – Snapdeal, Grofers, Practo, Oyo, Ola – attempting an Uber-for-X (an Uber-like app for any commodity) play on traditionally fragmented and informal markets. This resulted in a wild race to acquire customers through deep discounts and uncomfortably large marketing spends without keeping basic transaction level economics in check. Speculative valuations soared and then finally crashed. Venture funding also contracted by 50% quarter-on-quarter. These unicorns may well soon turn into unicorpses.
The media and financing craze actually perverted incentives to entrepreneurs, and fuelled these unsustainable business models. While Indian startups borrowed foreign business models, they forgot to buttress them with defensible data strategies and lock-in long-term value. First, vast amounts of data inflow across different channels has not been captured. Second, whatever data is captured has not been synthesised and analysed to draw actionable insights. Third, such insights have not translated to measurable significant business improvements.
Take e-commerce as an example. Over $5bn worth of funds injected into the top two domestic players has generated about 50 million online shoppers, who continue to flit across sites hunting for the best deal. While this investment has increased consumer access, it has not been able to reinvent e-commerce as we know it. These platforms have neither seen any sustained increase in conversion rates, nor have they reinvented buying experiences by purchase category. Even today, a consumer faces the same generic experience whether buying a pack of diapers, high-end fashion, or a washing machine!
In a roundabout way, this is actually good news for the 4,300-strong Indian startup system as the hunt for real value has begun. A more selective funding climate now has restored proper incentives and thrown better light on Bangalore and India’s true advantages.