Thae U.S. holds an enviable lead in pushing artificial-intelligence technology out of labs and into real-world applications. Thank companies like Alphabet(GOOGL), Facebook (FB) and Apple (AAPL) for that.
But China’s government and technology elites aim to overtake the U.S. in AI by 2030 — or so they proclaimed in July at a Beijing political gathering.
Good luck with that.
Yes, China has many strengths as it sets out for worldwide dominance in AI technology. Its internet giants Baidu (BIDU),Alibaba Group Holdings (BABA) and Tencent Holdings (TCEHY) are also pouring money into AI research and hiring top scientists.
China’s huge population will generate massive raw data to train AI systems in how to make predictions. So there’s good reason to think China will make breakthroughs in developing computer algorithms — the software programs that aim to replicate the human ability to learn, reason and make decisions.
China also has a major weakness: a semiconductor industry that still lags the U.S. in making high-end electronic processors. Chinese companies buy AI chips mainly from Nvidia (NVDA), based in Santa Clara, Calif. Intel (INTC), the dominant supplier of brainy chips for personal computers, is pushing fast into AI.
“Most AI-focused chips are developed by U.S. companies, and it will be a long slog for China to catch up,” said David Kanter, head of chip industry consultant Real World Insights.
UBS forecasts that the AI chip market will boom to $35 billion by 2021, up from roughly $6 billion in 2016. But AI stakes are higher than mere chip sales.
AI technology is expected to transform economies. Artificial intelligence has been called a “winner take all” technology, meaning that companies and countries that gain an edge will build upon that lead over time.
Russian President Vladimir Putin has said that whichever country leads in AI will dominate global affairs.
“Artificial intelligence is the future, not only for Russia, but for all humankind,” said Putin. “It comes with colossal opportunities, but also threats that are difficult to predict. Whoever becomes the leader in this sphere will become the ruler of the world.”
Putin’s problem: Russia doesn’t have high-end commercial chips, even though its military electronics are world-class.
All AI software needs computing power to find patterns and make inferences from large quantities of data. The race is on to build AI chips for data centers, self-driving cars, robotics, smartphones, drones and other devices.
“The war for AI architecture leadership is the biggest battle of our lifetime,” said Gary Dickerson, CEO of chip equipment maker Applied Materials (AMAT), in an interview. He says new, specialized processors for AI software are coming. Applied Materials sells to both U.S. and Chinese chipmakers.
Right now, the U.S. is ahead in startup AI companies and commercial deployment. U.S. Fortune 500 companies are pushing AI analytical tools into finance, health care, energy, agriculture, cybersecurity and elsewhere.
Tech giants Apple, Google-parent Alphabet, Facebook and Microsoft have forged ahead in applying AI software to speech recognition, internet search, and classifying images. Amazon.com‘s (AMZN) AI prowess spans cloud-computing services and voice-activated home digital assistants.
Rise Of China In AI
China does plan to catch up. Just not right away.
“They are in it for the long run,” said Chirag Dekate, a high-performance-computing analyst at research firm Gartner. “The U.S. is driving AI innovation across the spectrum, in software and hardware. Early-use cases, early adopters — it’s happening more in the U.S. than any geography in the world. But China is looking at it from a marathon perspective.”
China’s government in July unveiled a three-step development plan to steadily build up AI capabilities through 2020 and 2025 and to lead the world by 2030.
Building up AI engineering talent is key to the plan. Chinese scholars already are churning out AI research papers at a faster pace. Patents have soared in robotics and other areas. Chinese internet companies have set up AI research in Silicon Valley. Meanwhile, Baidu and other companies are paying top salaries to snatch AI scientists.
And yes, high on the Chinese government’s to-do list by 2030 is developing high-end AI chips.
“Computing power is part of the basic infrastructure underlying AI and of significant strategic importance,” said a report on China by Vertex Holdings in July.
“Stronger control over the supply of core technologies can potentially improve China’s future ability to deploy AI systems more widely,” the report added. Vertex is part of Temasek Holdings, the Singapore state investment firm.
U.S. Chip Stocks’ AI Approaches
Just a few years ago, AI research projects required hundreds of computers networked together to process complex software. One key to AI’s spread in the global economy, says a CBI Insights report, is a new wave of AI chips.
Nvidia has emerged as the early leader in AI chips. Nvidia’s edge is that its PC gaming processors can be scaled up to handle AI software, thanks to their “parallel processing” circuitry that can handle complex multiple tasks.
Baidu has been Nvidia’s partner in self-driving cars. In September, Nvidia said e-commerce giant Alibaba Group, Tencent and others will use its AI chips in cloud-computing data centers.
Intel has made several AI-related acquisitions. Intel says it’s working with Facebook to develop an AI chip for cloud-computing services.
In the U.S., it’s not just traditional semiconductor firms that are developing AI chips. Google’s strategy is melding AI software and chip technology. Google’s TensorFlow data-center software runs on its own “TPU” chips.
Apple’s new iPhone X features its own machine-learning optimized A11 bionic chip. The AI chip supports facial recognition as a way to unlock phone screens. The “bionic” processor is just the start of Apple’s push into AI chips, analysts say.
U.S. Government Protections
The U.S. government, meanwhile, has been vigilant in protecting high-end chip technology.
The Trump administration in September blocked the sale of Lattice Semiconductor(LSCC) to a Chinese-backed investor on national-security concerns. Lattice owns programmable software technology that offers an alternative way of building AI chips.
The Obama administration in 2015 barred Intel, Nvidia and Advanced Micro Devices (AMD) from selling high-end supercomputer chips to the Chinese government over concerns they’d be channeled to military systems.
China’s scientists responded a year later by unveiling Sunway TaihuLight, which broke records as the world’s fastest supercomputer and contained no U.S. intellectual property. Some observers view Sunway’s accomplishment as a sign China could be competitive in AI chips.
China Investment In Artificial Intelligence
China is still a far way, though, from producing AI chips for the mass market. One startup, Cambricon, has attracted funding from Alibaba, Lenovo Capital and IT firm iFlytek. Singapore’s Vertex is among the backers of another startup, Horizon Robotics.
Huawei, one of China’s top tech companies, also has AI chip projects underway.
While China plans a big push into AI, it also plans to up spending on its semiconductor industry. China has put $20 billion into a new chip industry project and could spend as much as $150 billion, according to a U.S. government estimate.
China still buys most high-end chips from foreign suppliers, even though it has twisted the arms of U.S. and European companies since the 1990s to share technology through joint ventures. Intel, Advanced Micro Devices and SoftBank-owned ARM dominate in microprocessor technology, where China has little intellectual property.
“The progress of building an indigenous semiconductor industry has been slow in China,” noted a Goldman Sachs report on artificial intelligence released in September. “We expect lower foreign dependency over time.”
Google reportedly has been trying to interest China’s tech companies in its TensorFlow AI software tools, which make it easier to develop apps. But it’s not clear if those talks have involved its AI chips as well. Google’s cloud-computing service rents access to its TPU chips optimized for TensorFlow.
Venture-capital investment in AI chip startups is soaring, says CBI Insights. While Nvidia’s GPU chips have grabbed an early lead, startup AI companies are focused on developing chips designed from scratch to crunch artificial-intelligence software. They include U.K.-based Graphcore, KnuEdge and Cerebras Systems.