April 23, 2018
PUBLISHED BY Tiernan Ray
“Have you ever ridden in a self-driving car?” Tom Siebel asks me, over a lunch of grilled salmon and tuna Niçoise in midtown Manhattan.
Siebel has never ridden in one himself. “I wouldn’t get in one,” he says, shaking his head about the prospect of autonomous vehicles. That is despite the fact that Siebel has based the future of his company, C3 IoT, on artificial intelligence and machine learning.
Siebel is a legend in the software industry, as described in this space last August, having sold his company, Siebel Systems, to Larry Ellison’s Oracle (ORCL) in 2006 for $5.85 billion. C3 IoT is focused on using AI for things such as predicting failure of oil rigs and jets and saving companies money by so doing.
“There are hormones,” Siebel observes of human beings. The current attempts by computers to model human thought or behavior, including what they do behind the wheel hurtling down the highway, is based very narrowly on just a limited set of synaptic interactions, he posits. AI ignores all the other things that happen in people, including hormonal interactions, the whole web of human biology, for that matter. For that reason, autonomous driving will fail, he anticipates.
“I don’t buy it,” he says of self-driving initiatives, with a smile. That includes not just Alphabet’s (GOOGL) Google’s Waymo business, but also the efforts of Nvidia (NVDA), the chip maker that has been one of the most aggressive advocates of self-driving vehicles, which it hopes will sell lots of Nvidia car chips. I point out that Jensen Huang, founder and CEO of Nvidia, has confidently predicted lots of self-driving vehicles will be rolling into the lot in 2020 or 2021.
“Jensen is very, very smart, I mean, he’s just incredibly smart,” says Siebel. “I could be wrong,” he allows, “I just don’t buy it,” he repeats.
AI’s Mixed Bag
Siebel’s view is a mix of enthusiasm and dread about AI.
He thinks it will bring incredible benefits to society, as big as the industrial revolution. He also believes it will bring very grave costs just like the steam engine. The rise of communism, child slave labor, and the displacement of populations are things he mentions from the last century as examples of the kind of terrible punishment humans could be in for without proper regulation of AI
“There’s a lot of good, and a lot of bad” in AI, Siebel says. “The economic and social benefits will be staggering, but the human costs are also going to be significant, we need to anticipate that.”
The plus side is embodied in his own company’s success. Siebel updated me that C3 IoT, which buys tons and tons of Huang’s GPU chips to run its AI algorithms, has signed up John Deere (DE), 3M (MMM) and Royal DutchShell (RDSA), and various branches of the U.S. Government such as the U.S. Air Force, as customers. Without disclosing revenue figures, he noted C3 IoT’s bookings more than doubled in the fiscal year ended in January, after rising 200% the year before. and 600% in 2016. Bookings will more than double again this year, he expects.
Revenue rose by more than 50% last year. (When we last met, in August, Siebel remarked revenue was about $100 million annually.)
The things his software does with algorithms and the sensors placed on oil rigs and aircraft are to help predict the failure of assets days or weeks in advance. Sensors have been proliferating in industrial equipment, but conventional technology will use them only to tell you when something is actually failing–it can’t predict. “When you look the number of pumps and compressors in an offshore oil rig, the complexity is enormous,” he says. “Any one of those components could cause catastrophic failure.”
The same goes for things such as the Air Force’s “C-5 Galaxy” transport aircraft that move tanks. Anticipating such failure is his company’s business. His company uses tons of Nvidia chips, by way of the cloud computing services it uses, to perform all that AI, as no one has come close to what Nvidia’s achieved, he argues. (He’s not counting out Intel (INTC), however.)
Siebel has formed partnerships with Microsoft (MSFT), after long being mostly reliant on Amazon.com‘s (AMZN) AWS. He indicates that C3 IoT is increasingly looking at the “multi-cloud” opportunity for hosting and selling its AI, not being captive to just one cloud-computing operation.
C3 IoT has raised over $300 million to date in financing, including a round of $100 million in January, of which TPG was the lead investor. He notes the company has “substantial cash reserves” and doesn’t expect to burn any cash this year.
Read the full article on Barron’s here.