We are currently facing another gold rush in AI services . Startups are investing billions in AI startups in every industry and business performance. Google, Amazon, Microsoft, and IBM invested more than $ 20 billion in AI in 2016. Corporations are scrambling to look over the shoulders of startups and realize the productivity benefits of AI over their competitors. China is putting its considerable weight behind AI and the European Union is talking about $ 22 billion in AI investment because it fears losing ground to China and the US.
AI solutions are everywhere. From 3.5 billion searches every day on Google to the new Apple iPhone X, which uses Amazon Alexa for facial recognition, it answers our questions beautifully. How AI can help diagnose diseases, banks better predict customer credit risks, farmers predict crop yields, marketer’s target and retain customers, and manufacturers improve quality control. There are think vessels committed to studying the physical, cyber and political risks of AI.
So who makes money in AI?
AI and machine learning are ubiquitous and woven into the fabric of society. Like any gold rush, who will find gold? Will it be brave, low and big? Or can snappy upstarts hold their nuggets? Do those who offer picks and shovel make more money? Who pays for the dust?
Where is value being created with AI?
As I begin to think about who is going to make money in AI, I end up with seven questions. (1) Chip manufacturers, (2) platform and infrastructure providers, (3) launching models and algorithm providers.
1. Who got the best AI chips and hardware?
Although the price of computing power has fallen tremendously, demand is growing even faster. AI and machine learning have an appalling and insatiable appetite with its huge datasets and its trillion vector and matrix calculations. Bring on the chips.
Over the last two years NVidia’s stock has grown by 1500%, and their graphical processing unit (GPU) chips have historically been used for machine learning to provide beautiful high speed flying games graphics. Google recently launched its second generation of Tensor Processing Units (TPUs). And Microsoft is building its brainwave AI machine learning chips. At the same time, startups like Graph Core, which raised more than $ 110 million, are looking to enter the market. Current chip providers such as IBM, Intel, Qualcomm and AMD are not standing still. Rumor has it that Facebook is also building a team to design its own AI chips. With Cambrian Technology announcing its first cloud AI chip this past week, the Chinese are emerging as serious chip players.
2. Who has got the best infrastructure and platform cloud for AI?
The AI race is also taking place in the cloud. Amazon realized early on that startups would rent rather than buy computers and software. That is why it launched Amazon Web Services (AWS) in 2006. AI today demands so much compute power, companies are leaning towards the cloud to hire hardware through infrastructure (IaaS) and platform one service (PaaS) offerings.
Amazon tops cloud services, but their stories include Microsoft, IBM, Google, and Alibaba.
The fight between tech giants continues. Microsoft is offering their hybrid public and private Azure cloud service that has over a million computers. In the past few weeks, they have announced that their Brainwave hardware solutions will dramatically accelerate machine learning as they improve their Bing search engine performance by a factor of ten. Google is rushing to play catchup with its Google Cloud offering. We are seeing Chinese Alibaba start taking a global stake.
Large cloud players are racing to make sure they are positioned for the huge demand driven by AI.
Amazon — Microsoft — Google and IBM continue to do it. And look for the heavily scaled cloud players from China. The big picks and shovel guys win again.
3. Who has got the best enabled algorithms?
Today Google attracts some of the world’s best AI companies, spends small country-sized GDP budgets on R&D, and sits on the best datasets generated by billions of users of their services. AI also powers Google’s search, autonomous vehicles, speech recognition, intelligent reasoning, mass search and its work on drug discovery and disease dialysis.
And the incredible AI machine learning software and algorithms that power all of Google’s AI functionality — Tensor Flow — are now being offered for free. Yes for free! Tensor Flow is an open source software project now available to the world. And why are they doing this? As Google Brain head Jeff Dean recently put it, 20 million companies in the world today can benefit from machine learning. With millions of companies making good use of class-free AI software, they need a lot of computing power. And who better serves it? Optimized for Google Cloud Tensor Flow and related AI services. Once you rely on their software and their cloud, you will be a very sticky customer for many years to come. No wonder that this is a brutal race for the dominance of the global AI algorithm with Amazon — Microsoft — IBM.