'Data is the epicenter of the AI revolution': How AI is disrupting the data center and enterprise IT markets

Samantha Lee/Business Insider

• The convergence of data and cloud technology has enabled businesses of all sizes to use AI for automation, better customer information, and online transactions.
• Facebook uses AI tools Sapienz and Infer to debug source code, while Google's AutoML helps develop new machine-learning software.
• This article includes an overview of AI in enterprise technology, plus the top three trends to watch, and how Google and Facebook use AI on their products.
• Read how AI is transforming health, transportation, investing, and more in other articles from our special report, How AI is Changing Everything.

Artificial intelligence has accentuated the ongoing disruption of two huge markets: the data center and enterprise IT.

The rise of cloud computing over the past decade has radically changed the way businesses - from big corporations to mom and pop shops - access and use computing resources.

The cloud made it possible for them to set up and run their computing infrastructure, allowing them to downsize, if not totally abandon on-premise data centers, dramatically cutting IT costs.

AI is helping speed up and enhance the benefits of that transition. Traditional tech powerhouses have played an important role in this change.

Ruchir Puri, IBM's chief scientist, said AI's rise in enterprise IT was due largely to the explosion of data - online transactions, wide-ranging customer information, social media posts - that businesses could now collect, process and draw valuable insights to use for making tactical and strategic decisions.

"Data is the epicenter of the AI revolution," he told Business Insider. "A lot more data became available. The whole data infrastructure became more prevalent. Then a lot more compute available which was needed to deal with that data."

In fact, AI's impact extends to the basic component of modern computing: the chips used to power data centers and the cloud.

AI triggered a push for more powerful processors that could handle the more intensive computing needs of AI systems, especially the fast-rising trend called deep learning, which is based on neural computing networks that mimic the way the human brain works.

More powerful AI-enabled data centers and cloud platforms paved the way for more powerful, more sophisticated business software applications.

AI is making it possible for businesses to automate key parts of their operations, from sales, customer relations, managing inventory to long-term planning. AI software applications are enabling business leaders to make faster and more insightful decisions, helping them save time and money.

Enterprises worldwide are expected to spend $35.8 billion on AI-based systems in 2019, said a recent report from the analyst firm IDC. AI is spreading so rapidly that by 2022, three-fourths of IT operations will be handled by "AI or analytics-driven automation," cutting operational expenses by more than 25%, another IDC report also said.

"It's about pattern detection. It's about autonomous decision-making," Juergen Lindner, an Oracle executive focused on the tech giant's cloud software business, told Business Insider.

AI is making it possible for business leaders to make far-reaching decisions. It is "elevating their decision-making to an unprecedented level," Lindner said.

Facebook

Top 3 opportunities in AI for enterprise tech

Cybersecurity: Growing

Software makers have been helping businesses defend against hacks and other network threats for decades. AI is creating new cybersecurity capabilities on steroids, offering faster real-time protection and pinpointing potential risks long before they hit. It's a particularly serious need as more companies take their networks to the cloud. Crowdstrike, an AI cloud security startup, went public last month after raising $600 million.

Sales and Customer Relations: Growing

Cloud platforms, such as Salesforce, already let sales reps and managers keep track of customer leads and accounts. AI is dramatically expanding what they could do with all that data, letting them quickly figure out which leads are most promising, which sales rep is best suited for a customer, or how best to engage with a client.

Programming: Nascent

Developers at Facebook and Google have been using artificial intelligence to help them be more productive. At Facebook, developers use the AI tools Sapienz and Infer to debug and test code. At Google, developers use AutoML to automatically create artificial intelligence algorithms for products like Google Photos.

"I think this is going to be one of the most transformative but also one of the most rapid changes we're going to have," Rajen Sheth, vice president of product management at Google Cloud told Business Insider. "Now we have a lot of the tools that set that up for success."

AI tools Facebook and Google use to debug code and detect malware

At Facebook, engineers frequently use two AI tools - called Infer and Sapienz - to test and debug their code. Alexander Mols, software engineer on the Sapienz team at Facebook, says that before, testing code was a tedious, time-consuming process that would still result in errors.

Now, it uses Infer to scan the source code for errors, while Sapienz is a simulator that mimics the way an actual human would use the app. Mols says these two tools "naturally complement each other" to find errors, bugs, and glitches.

Infer came out of Facebook's acquisition of the startup Monoidics in 2013, while Sapienz came out of its acquisition of Majicke in 2017. Both these tools are used for Facebook, Messenger, Instagram, and Workplace, while WhatsApp makes use of Infer.

"It takes a lot of work away from the engineers," Mols said. "Instead of them having to spend time writing tests for the code, the algorithms take care of this for them."

Still, Mols says, there are things Sapienz and Infer can improve on. Right now, the team is working on improving the capabilities of these tools, such as adding automatic fixing and better explanation of the bugs it finds.

"We could work on better ways to document and explain these issues to engineers," Mols said.

Google is taking a slightly different approach: The search giant has developed a way to make it easier to use machine learning - the technology where a computer algorithm can learn from data, identify patterns, and make predictions without a human explicitly telling it what to do.

Google's tool is called AutoML, and it essentially helps programmers - both internally at the tech titan, and its Google Cloud customers - use machine learning software to train machine learning software.

It might sound meta, but the idea is that developers - even those who don't know much about working with machine learning - can bring their data sets to Google Cloud. AutoML will automatically create a machine learning model, which can be used for making predictions, identifying images and videos, or understanding text.

"What we found with that is we can use the latest deep learning technologies to give people models that are highly accurate to their dataset," Rajen Sheth, vice president of product management at Google Cloud, told Business Insider. "They don't have to have machine learning expertise to do this."

Within Google, AutoML is used for tasks like recognizing and aggregating photos in Google Photos, automatically composing replies in Gmail, and detecting malware.

"What we see within Google is that within a period of four years, we've gone from very little AI within products in Google to using it in every single product," Sheth said. "Now almost every developer at Google has access to machine learning."

Sheth says these tools make machine learning easier to more developers, helping them take advantage of the rising popularity of the technology. Now, Sheth says, the challenge will be to help those programmers stay on top of the latest trends with the technology.

"We're going from a world where machine learning has been very, very new, to now a world where there are many more tools and so many more people can access it," Sheth said. "Over the next decade, every single business will be radically transformed by AI. We need to transform every developer to use this in the right way for their businesses."

The Takeaway

Rick Wilking/Reuters

Jensen Huang, CEO of Nvidia, shows the old (R) and new computers for autonomous vehicles at his keynote address at CES in Las Vegas, Nevada, U.S. January 7, 2018.

"AI is the future of software. Software is the language of automation. It's very clear that AI is going to impact every industry."

-Jensen Huang, CEO of Nvidia to VentureBeat