AI startups abound from Silicon Valley to London to Shanghai. However, like with any gold rush, a few lucky people will strike gold while others will be dissatisfied. To be successful, AI firms must transcend the commercial divide from technology to the enterprise. And while embracing appropriate AI methods, learn not to move too quickly and break things. AI is in full-fledged gold rush mode. Every day, we read about AI firms raising massive sums of money to prospect AI gold veins. Money is pouring into these new frontiers. In theUnited States, venture capital funding for AI increased by 72 percent year on year to a stunning $9.3 billion in 2018.
For example, Dataminr, aNew York-based AI and machine learning business that analyzes news and information in real-time, raised $392 million in 2018. Last year, AI-powered computer security and management firms in Silicon Valley each raised more than$100 million. Pony.ai also collected $102 million to develop its latest self-driving technology. However, the boundaries of the Wild West have now opened up in the East. China, to be specific.SenseTime, which focuses on revolutionary computer vision and deep learning, secured $600 million in two tranches in 2018.
It now claims to be the most valuable AI startup in the world. UBTECH, which describes itself as an AI and humanoid robotics business, raised $820 million in 2018. Face++, the market leader in facial recognition, has raised over $600 million and is currently seeking $500 million. And iFlytek, which claims to control 70% of the Chinese speech recognition industry, aims to raise $565 million. These are huge statistics, and we can expect much bigger ones in the future. A few AI breakout stars in Europe, such as Dataiku, a French AI tools firm that assists enterprises in moving AI from experimental to production. They raised more than $100 million in 2018, but big money comes from American venture capital firms like NewYork's FirstMark.
The latest London AI Mayor's study revealed over 650 local AI startups, more than doubling Berlin and Paris combined. Many of these businesses are fresh startups focusing on the insurance, financial, and legal sectors in high demand. Money has flooded these eager young companies, but funding is typically in the millions and a few tens of millions at this stage, as opposed to hundreds of millions in the US andChina.
However, as with any gold rush, many prospectors are making exaggerated claims. When you walk into any company pitch event or AI meetup — the modern-day frontier saloons — you'll hear startup owners braying loudly about how the newest profound learning discoveries power their firm. However, digging beneath the surface in the cold light of day appears to provide a lot of fool's gold.
According to a new AI analysis by London MMC Ventures, more than 40% ofEurope's 2,830 so-called AI businesses are not using AI. There are even allegations that AI tech behemoths like Microsoft and Google are not always open about relying on people to power their "AI solutions." Appen, an Australian startup, validates search results globally with a global staff of over 1,000,000 flexible workers. It's a genuine army of individuals painstakingly reviewing search results. Not what we would expect from these global AI leaders who promote total automation.
We define AI startups as those that either (1) would not exist if AI modern technologies, such as deep neural networks, did not exist — it is fundamental to their existence — or (2) provide AI infrastructure and tools, such as AI specialist hardware, cloud services for AI applications, or tools to accelerate the implementation of AI solutions.
The AI playing field is vast, with AI creators (those who create AI technologies) and takers (who consume AI technologies) and extract value. In part one of this series, the established seven-layer value chain for determining who will profit from AI are:
(1) AI chip and hardware makers looking to power all of the AI applications that will be woven into the fabric of organizations big and small around the world;
(2) cloud platform and infrastructure providers who will host the AI applications;
(3) AI algorithmic and cognitive services building block makers who provide the visual recognition, speech, and deep machine learning predictive models to power AI applications;
(4) enterprise solution providers whose wares will be woven into the fabric of organizations big and small around the world;
(5) Industry vertical solution providers looking to use AI to power companies in sectors ranging from healthcare to finance;
(6) corporate AI adopters looking to increase revenues, drive efficiencies, and deepen their insights; and
(7) nation-states looking to embed AI into their national strategies and become AI-enabled countries.
While AI startups seek to provide novel processors, cloud services, and algorithms, this segment of theAI value chain is dominated by deep-pocketed technology behemoths such asGoogle, Microsoft, and Amazon. They've become the gold rush's picks and shovels. The giants want to ensure that whichever AI firm is searching for gold is powered by their AI hardware, cloud, and algorithmic solutions.
It costs a fortune to design, build, and distribute hardware chips. Therefore only a few businesses can thrive. Graphcore, situated in the United Kingdom, has raised more than$110 million in developing machine learning-optimized semiconductors. However, they compete in a crowded field like Google, Facebook, and Microsoft to launch their AI-optimized CPUs. Google Cloud, for example, has introduced Cloud Tensor Processing Units, which enable its cloud services. In addition, chip incumbents such as IBM Investigations and Intel are not sitting still as they develop their AI-optimized CPUs. Cambrian, a ChineseAI chip manufacturer, powers hundreds of millions of Chinese mobile phones.