SAN FRANCISCO — As the oracles of Silicon Valley debate whether the latest tech boom is sliding
toward bust, there is already talk about what will drive the industry’s next growth spurt.
The way we use computing is changing, toward a boom (and, if history is any guide, a bubble) in
collecting oceans of data in so-called cloud computing centers, then analyzing the information to build
The terms most often associated with this are “machine learning” and “artificial intelligence,” or “A.I.”
And the creations spawned by this market could affect things ranging from globe-spanning computer
systems to how you pay at the cafeteria.
“There is going to be a boom for design companies, because there’s going to be so much information
people have to work through quickly,” said Diane B. Greene, the head of Google Compute Engine, one of
the companies hoping to steer an A.I. boom. “Just teaching companies how to use A.I. will be a big
This kind of change is what keeps Silicon Valley going. When personal computers displaced mainframe
computers, it opened the door not just for Apple, but for companies making PC software for business,
games and publishing. In the networking and Internet revolutions, venture capitalists invested in these
new computing styles, and another generation of companies was born.
Over the last decade, smartphones, social networks and cloud computing have moved from feeding the
growth of companies like Facebook and Twitter, leapfrogging to Uber, Airbnb and others that have used
the phones, personal rating systems and powerful remote computers in the cloud to create their own
Believe it or not, that stuff may be heading for the rearview mirror already. The tech industry’s new
architecture is based not just on the giant public computing clouds of Google, Microsoft and Amazon,
but also on their A.I. capabilities. These clouds create more efficient and supple use of computing
resources, available for rent. Smaller clouds used in corporate systems are designed to connect to them.
The A.I. resources Ms. Greene is opening up at Google are remarkable. Google’s autocomplete feature
that most of us use when doing a search can instantaneously touch 500 computers in several locations as
it guesses what we are looking for. Services like Maps and Photos have over a billion users, sorting places
and faces by computer. Gmail sifts through 1.4 petabytes of data, or roughly two billion books’ worth of
information, every day.
Handling all that, plus tasks like language translation and speech recognition, Google has amassed a
wealth of analysis technology that it can offer to customers. Urs Hölzle, Ms. Greene’s chief of technical
infrastructure, predicts that the business of renting out machines and software will eventually surpass
Google advertising. In 2015, ad profits were $16.4 billion.
“In the ’80s, it was spreadsheets,” said Andreas Bechtolsheim, a noted computer design expert who was
Google’s first investor. “Now it’s what you can do with machine learning.”
He added: “Better maps and photos is just the start. It’s going to be in life sciences, automobiles,
A number of start-ups are already aimed at the new architecture. A Mountain View, Calif., outfit called
Mashgin uses “computer vision” to automate retail checkout. Up Highway 101 in San Mateo, a company
called Alluxio is creating ways to make cloud-based A.I. work better. Last week, a San Francisco
company called Mesosphere, which makes a way to operate among various corporate and public clouds,
raised $73.5 million.
Microsoft and Amazon are racing Google to dominate the new architecture.
This week, Microsoft will kick off a conference in San Francisco that is expected to focus on ways
machine-based intelligence can be used to analyze, among other things, “the Microsoft graph,” or all the
data companies already have in the Microsoft products they’ve owned for decades.
Amazon last year announced its own machine-learning services, and it is amassing its own large
repository of corporate data.
Hewlett Packard Enterprise, an older company struggling to find its way in the new landscape, was one
of the investors in Mesosphere.
“When you are building predictive data, you don’t know what you are going to need next,” said William
Hilf, a senior vice president at H.P.E. “If someone makes a bet in machine learning on Microsoft or
Google, they may need to come down to their old data systems, too. We are building platforms to bridge
among all of them.”
To Ms. Greene, all of the activity so far, along with the size and sophistication of computing, is small
compared to what will happen when the world’s biggest businesses start leaning on the new A.I.
“We may build an A.I. system to figure out all the ways businesses can use this,” she joked. “The
relationship between big companies and deep machine intelligence is just starting.”