- Konstantine Buhler, a partner at Sequoia Capital, believes “cloud is going to become AI.”
- Buhler insists that AI is not “magic,” and that it should be demystified and measured.
- Buhler says companies can bake AI into their processes “horizontally.”
- This article is part of a series about cloud technology called At Cloud Speed.
If you ask Sequoia Capital partner and early-stage investor Konstantine Buhler about the role of artificial intelligence in cloud computing, his answer is unequivocal: “Cloud is going to become AI,” he told Insider. “I mean, all of the cloud will be based on AI.”
Snowflake’s $3.4 billion initial public offering and DataBricks’ $1 billion funding round over the past year suggest big things ahead for AI in the cloud, and the industry is estimated at $40 billion and climbing. Major platforms like Amazon’s AWS, Microsoft Azure, and Google Cloud – as well as a host of startups – sell cloud-based tools and services for data labeling, automation, natural language processing, image recognition, and more, making it more affordable than ever before for firms to dabble in AI.
Buhler, who has a master’s degree in artificial intelligence engineering from Stanford, revels in AI’s contributions, but also insists that the sector be demystified, and basic business fundamentals applied to it.
His investments include CaptivateIQ, which automates business commissions, and Verkada, a security camera company that uses AI to recognize information like license plate numbers. Sequoia in general is an investor in some of the biggest names in AI, including Snowflake and Nvidia.
“This next wave of enterprise and consumer technologies will all need AI built in,” Buhler said. “That’s going to be the standard going forward.”
AI’s ubiquity in the future is the first of a few basic lessons Buhler believes everyone should understand about AI’s impact over the next decade in the cloud. Here are the rest:
AI is not magic – it’s math
There is an (unwarranted) aura around artificial intelligence that ascribes to it supernatural brilliance.
“It seems complicated – it seems like magic of some sort, so people get intimidated and awed by it,” Buhler said. “Artificial intelligence is just more and more mathematical computations done rapidly, which at some point, for a moment, seems ‘magical.’ But it never is.”
Ordinary people should ask to understand it, because it impacts their lives. If you talk to Apple’s Siri or Amazon’s Alexa, you are conversing with AI. If your cat hops aboard a Roomba vacuum, both of you can appreciate how it “learns” to avoid objects in its path. On the other hand, a red-light camera that zooms in to read your license plate when you go through intersections late and automatically fines you might not be such a welcome innovation.
AI should learn from the internet revolution
Buhler believes that AI is at a similar inflection point as what the internet revolution experienced 20 years ago: “Let’s learn a lesson from the dot.com boom,” when many over-valued companies imploded as they failed to materialize as real companies, Buhler said: “Everybody had that mentality of, ‘let’s stick internet on this thing.'”
While cloud-based tools allow companies to spin up AI models with relative ease, not every problem needs to be solved with these kinds of algorithms.
The business case must always be there – with the customer centered – or AI will not be practical.
“When you build an artificial intelligence model, it is not about the AI: It is about the customer,” Buhler said. “The internet was a communication revolution, and AI is a computation revolution. This is a new mechanism to serve people, and you have to understand their needs, or you’re going to spend years building the wrong thing.”
Every company has a ‘horizontal’ AI opportunity
Buhler believes every company can bake AI into their business using the same basic “horizontal stack,” or processes that take raw data and turn it into actionable intelligence that can be used in different ways across business units. Buhler says companies like Databricks, Dataiku, DataRobot, and Domino Data Lab (“they all start with D for some reason”) help enterprises do this.
Horizontal data processes can include data preparation (sorting text from image files, for instance), data labeling, data storage, creating algorithms that process the data, and, finally, applying the algorithms to specific business processes to help guide decision making.
“It should be laid out that simply,” he says. That process “is all about enabling enterprises to bake artificial intelligence directly into their systems.”
AI startups can also focus on verticals
Buhler says there are also AI startups that are providing products tailored to more specific business needs. Gong, for example, helps salespeople evaluate opportunities, while competitor Chorus turns sales conversations into data. In the financial world, the startup Vise automates investment management, while in the legal world, Ironclad helps attorneys build contracts faster. Gong, Vise, and Chorus are Sequoia portfolio companies.
The key in picking great AI startups, Buhler says, is being able to measure how a company is helping its customers: “It has to be a real business with outputs that can be quantified.”