How is Japan Doing in Artificial Intelligence, and What’s at Stake?
[English Version, for Japanese Version, click here]
The Japanese Economic Miracle
In 1980, Japan was on a trajectory to be the world’s #1 economy by size. Today, Japan is the world’s #3 economy by size. In July 2019, the Japan Center for Economic Research (JCER) released a long-term economic forecast through 2060 that shows:
The size of Japan’s economy basically flattening at around $5Trillion (in 2014 US$)
China rising to surpass the US and become the world’s largest economy around 2030, but dropping back below the US around 2060
India’s economy surpassing Japan before 2030, pushing Japan to #4 in economy size
Having worked in the high tech sector for most of my career, I’ve come to understand that forecasts beyond 2 to 3 years represent just one possible scenario due to the rapid pace of change. Therefore, my faith in the accuracy of a single 40 year forecast is, let’s say, close to zero. Of course JCER only presented one scenario out of many. I’m a huge believer in scenario-based planning coupled with management approaches that are able to adapt to rapid change. That’s a topic for another day - for now, there are still some key points to take away from JCER’s very long-term scenario:
Japan’s aging population challenge is a severe damper to Japan’s future growth.
China has an aging population challenge as well - the big impacts have yet to occur.
Intangible capital (for example software, IP, business models) are the source of growth with a shrinking / ageing population.
Shared digital data (in society) can increase Japan’s efficiency.
Globalization brings the benefits of international specialization, and enhances productivity through competition and cross-border information exchange.
Before we look forward from today, let’s look back at what drives growth, and how changes in productivity drive changes in global economic share, and also create substantial shifts in global power over time.
In this amazing graphic showing 2000 years of economy size progression, the 1800 years to the left of the white gap shows world powers GDP share being driven only by population differences. Output corresponded to what one person could do with their hands and simple tools. To the right of the white gap is the Industrial Revolution and beyond. Here a divergence between global populations and productivity began to occur, with the initial beneficiary being the UK, and later Germany and the United States. The US compounded its growth with immigration. Japan was the next phenomenon during the “Japanese Economic Miracle” of 1945-1990. The distinguishing characteristics of the Japanese economy during these years included: the cooperation of manufacturers, suppliers, distributors, and banks in closely knit groups called keiretsu; the powerful enterprise unions and shunto; good relations with government bureaucrats, and the guarantee of lifetime employment (shushin koyo) in big corporations and highly unionized blue-collar factories. Now we see China rising through a period of urban population growth combined with focused central government strategies including a focus on Artificial Intelligence.
Is Artificial Intelligence a Buzzword, or the Driver of the Next Wave of Productivity Gains?
Well, probably both. Artificial Intelligence (AI) is a huge buzzword in tech today. On IT company earnings calls, the term “Artificial Intelligence” surpassed “Cloud” in 2011, “Big Data” in 2016, and was mentioned more than twice as often as “Machine Learning” in 2017. Despite being over-hyped (and likely exaggerated in marketing as well as financial reporting), Artificial Intelligence is gradually fueling a new revolution, and includes machine learning techniques such as deep learning and neural networks, applications such as machine vision and natural language processing, and it enables and accelerates fields such as robotics, autonomous vehicles and new drug discovery. There’s certainly a race between the US and China for AI and quantum computing dominance, which has implications far beyond economics and into national/regional security. We’ll talk about the balance of power in Asia and Japan’s strategic situation in another blog.
Why is AI Important to Japan?
Where AI matters for Japan is that it can create a capital-labor hybrid, which transcends the current capacity of capital and labor, leading to higher levels of Gross Value Added (similar to GDP), and thus higher productivity on a per-person measure. From the Japanese government perspective, it’s an important issue because there’s good potential for automation of tasks to address workforce and healthcare challenges with an aging population. For Japanese companies, there’s commercial opportunity associated with globalizing the AI technologies and processes involved in Japan’s key export industries such as transportation finished goods and parts, integrated circuits, and machinery. There’s also a risk of losing global share in industries like automotive if they don’t stay competitive.
Who’s Driving AI Forward in Japan?
Companies are the main drivers of AI innovation. According to WIPO Technology Trends 2019: Artificial Intelligence, out of the top 30 patent applicant organizations, companies represent 26 while only four are universities or public research organizations. This pattern of company vs. university dominance applies across most AI techniques, applications, and fields, demonstrating that while universities may lay the foundations, it's a country's companies that are the primary drivers of innovation. The Internet of Things (IoT), Big Data, and greater computational power mean that AI applications are more feasible today than they were at the time of the last AI wave in the 1980’s. In an effort to be more prepared for this, Japan aims to produce 250,000 AI experts a year from their universities and technical schools. We’re not talking about PhD level creators of new machine learning or neural network algorithms, we’re talking about AI application developers leveraging sensors, data, and the latest AI methods. Teaching so many students to apply these skills for innovative results will require some creativity, and there’s certain to be a shortage of available teachers as well. In the near term it will also be up to Japan’s large global companies to pick up the pace of preparing their workforces for a more digital future, with a focus not just on basic research, but on application of technologies to solve real problems, and the approaches required to bring those solutions to market. This may require a healthy mix of overseas acquisitions as well.
Are Japanese Companies Leaders in AI?
It is hard to say definitively. The chart below shows that out of the top 26 companies filing AI-related patents, there are 12 from Japan, six from the United States, two from Germany, two from China, two from Korea, one from Finland, and one from The Netherlands.
Global commercial AI-related revenue would be a better measure, but revenues in emerging, hyped areas is very difficult to measure. Can we count on annual reports or analyst estimates? There are significant caveats. Legacy products including hardware are often reclassified under hot new trends, as we saw when the markets were frothing over “cloud”. Global companies like IBM are probably more aggressive at doing this than Japanese companies, which tend to be more humble. Making success even harder to determine, the definition of the hyped terms shift over time until stabilizing somewhere around the time that Gartner’s hype cycle becomes almost too crowded to read at the “peak of inflated expectations”. This was the case when NIST published a comprehensive definition of Cloud in 2011. I’m observing the same confusion in AI - see Gartner’s 2019 AI hype cycle and how many areas it encompasses today. The market as IDC defined it goes well beyond software and includes servers, storage, and AI-focused services.
Which Companies are Likely in the Lead with Applied AI?
In the US, AI is driven by software giants such as Microsoft and Alphabet/Google, as well as IBM. According to analyst firm IDC, IBM has the number one share of the $28B AI market with $2.6B in revenue in 2018. In China, AI has huge support from the government as well as technology leaders like Tencent, Baidu, and Alibaba. IDC shows that Inspur is the Chinese company with the greatest share, at $900M, and growing at 110% per year. Japanese companies with the most at stake are paying the most attention to AI - this might roughly correlate to AI patent filings, or investment/acquisition activity (as in Softbank’s case). Japanese companies like Toyota will certainly seek to highlight their progress, and change perceptions, with the Tokyo 2020 Olympics. I’ve already blogged about that here.
What Do These Companies Have to Lose?
For Japan’s largest multinational companies in our list of Top 30 AI patent filers, $500M of revenue outside of Japan could be at stake. Looking at the chart below based on JMNC Solutions analysis of D&B data, manufacturing and wholesale trade represent more than half of that global business.
With the exception of NTT, Manufacturing and Wholesale Trade are among the largest at-risk areas for Toyota, Sony, Hitachi, Panasonic, Mitsubishi, and Canon. Also consider that this chart only shows at-risk revenue for Japanese companies that filed a lot of AI patents, under the assumption that AI matters the most to them. However, Japan has many more manufacturing giants with overseas revenue such as Nissan, Honda, Bridgestone, Komatsu, and more. How much harder will it be to remain differentiated in global markets for companies that aren’t keeping pace with digital trends such as AI? Are they acquiring these technologies instead, or licensing them from partners? And beyond having competitive products based on leveraging new technologies, there’s still the challenge of doing business at global speeds in a highly competitive environment.
How Does it Play Out in the Longer Term for Japan’s Multinational Corporations?
Most Japanese Multinational Corporations will look to expand their footprint outside of Japan with acquisitions, as we’ve seen with deals like the Takeda acquisition of Shire in biopharmaceuticals, and Hitachi’s offer for the ABB power grids business. According to Refinitiv (formerly Thompson Reuters), Japanese M&A activity hit new highs in 2018. Meanwhile, on a completely different battlefront, the Japanese Government’s Society 5.0 initiative will help set the direction for how and why Japanese society should embrace new technologies faster. One area where this could come together nicely for Japan is robotics. I think that Japanese companies will continue to iterate on their robotics solutions domestically as the population declines. Innovations in robotics, leveraging AI, could become very competitive globally. Based on my own experience, I believe that in this field, as with any other, success depends on engaging deeply with talented local leadership in global subsidiaries to jointly develop the best hybrid management approach for a world that moves more quickly than it did during the "Japanese Economic Miracle".