I am Article Layout

Select your investor profile:

This content is only for the selected type of investor.

September 2017

Digital disruption: artificial intelligence

Artificial intelligence is one of the fastest-growing areas of digital technology, transforming our lives  and our investment portfolios.

Robots are taking over the world. No, not via an army of cyborgs, but through a more gentle and subtle infiltration of intelligent technology.

The latest devices from the likes of Amazon, Microsoft and Google already enable us to switch on the lights, check the weather forecast, consult our diaries, listen to music, boil the kettle or book a taxi – all by voice command.1 Soon, AI – the replication of actions which normally require human intelligence by computers – will be able to achieve much more, supported by investment from digital companies who now see machine-learning systems as a major source of future growth.

The AI market is already large. Yet it is forecast to get much, much bigger. It is expected to balloon to USD127 billion by 2025 from USD2 billion in 2015.2 The US and China are leading the way so far, with the bulk of the investment coming from high tech, telecom and financial services sectors within those countries. 

Enabler for all sectors

For the global economy that could mean trillions of dollars of extra growth, through both increased productivity and via a boost to consumption as people spend money on new or improved goods and services.

Growing fast
Global AI revenues
AI digital disruption thematics

Source: Bank of America Merrill Lynch

The AI revolution won't be confined to just one industry. It is a pervasive enabling technology that will boost the fortunes of some companies but cause the demise of some others across a wide range of sectors. In many cases, the difference between commercial success and failure could be how effectively and how quickly firms deploy AI. 

For the majority of businesses, using AI effectively boils down to being able to analyse data – a lot of data. The annual amount of data we produce is forecast to reach 163 trillion gigabytes in 2025 – ten times more than in 2016.3

Through machine learning, algorithms can use historical data to spot patterns and predict what might happen in the future. The next step up is deep learning, where computers learn from their mistakes and refine predictions with each new fragment of information. They are thus becoming increasingly adept at recognising images and speech, as well as at natural language processing (NLP) – understanding spoken and written words just as humans do and responding within context.

AI is already making its presence felt in industries as diverse as health, retail and finance. In healthcare, it is boosting the power of medical diagnostics, paving the way for personalised medicine. In transport, it is the key technology behind self-driving cars. It is also transforming finance: banks and investment advisers facing an existential threat from robo-advice. Retailers also ignore AI at their peril. Machine-learning can improve logistics and also allow for greater product customisation.  

If it's true that businesses can't afford to ignore AI, then it's also true that companies specialised in developing AI-related technology could see a big boost to their top and bottom lines. This translates into an investment opportunity, and is the reason why AI is one of the key investment themes of our Digital strategy.

Smaller, smarter chips

There are two direct ways for investors to access AI investments: hardware and software.

On the hardware side, semiconductors are a rich hunting ground. Deep learning tends to require a lot of processing power. That means demand for powerful graphics processors (GPUs) which can support parallel processing and thus enable computers to analyse and make use of vast quantities of data.

Driverless cars, for example, have a lot of learning to do: both on their immediate environment and on how to react to different situations. To better accommodate this, Tesla has recently upgraded the processing power of its autopilot system by 40 times thanks to new GPUs. Any further tech developments are likely to be met with strong appetite from manufacturers.

Given that most of GPU chip production is outsourced (with GPU companies mainly focusing on the design rather than manufacturing), this should lead to better growth prospects in the foundry and OSAT Outsourced Semiconductor Assembly & Test) industries.

The more sophisticated the AI machines get, the more memory they will likely require to function. Just one second of autonomous driving can generate as much 1GB of data, for example. All that information needs to be stored somehow, somewhere. That should support demand for memory chips, as well as for cloud storage solutions.

Overall – based on our conversations with key players in the semiconductor industry – we believe that AI could account for around 25 per cent of total semiconductor demand by 2020, from 10-15 per cent today.

Super-powered software

The AI-driven transformation in software is, arguably, progressing even faster. We believe that digital software companies will benefit the most from the AI era as they can extract recurring revenue from subscriptions as well as new products. This is in contrast to semiconductor companies, who will mostly profit from a one-time purchase cycle.

Companies such as Facebook, Baidu, Salesforce.com or Medidata have access to massive amount of consumer or enterprise data in their respective fields and are able to offer value-added services based on AI to their customers.4

The scope for what data-crunching AI software can do is virtually limitless, and companies are feeling the benefits already. Netflix, for example, estimates that it is preventing over USD1 billion of potential revenue loss per year from cancelled subscriptions by offering tailored search results and recommendations.5 Amazon managed to reduce its warehouse operating costs by at least a fifth by employing autonomous robots.

Only by embracing the robotic revolution can we stay ahead of the game. 

Other applications in the pipeline include using big data to predict which policyholders are most likely to make big insurance claims, digitising credit assessment for loan requests and sifting through hours of surveillance camera footage to help identify crime suspects in seconds.

Ultimately, the key to corporate success in the digital world will be the ability to harness data and turn it into business opportunities. For investors, one of the most direct ways to tap into AI is by identifying the tech specialists who are delivering the best hardware and software. Only by embracing the robotic revolution can we stay ahead of the game.