Follow @BSnyderSF
At a time when companies spend
billions on patent wars and frivolous acquisitions, IBM makes money the
old-fashioned way: through R&D
Like millions of others, I made a point
of staying home a few nights in February 2011 to watch a computer
challenge the world's best "Jeopardy" players. IBM's Watson won, of
course. End of story? Just a stunt? Not at all. After about five years
of development and millions in R&D spending, IBM is taking its first
steps to bring Watson out of the lab (and TV studio) and make money
from it.
IBM, which made Watson a unit of its
very profitable software group, has already landed major clients in
health care and financial services (two were announced in March), with
more in the offing.
Sure, Watson has an obvious cool
factor. Haven't we been waiting since "2001: A Spce Odyssey" for a
computer that could really understand human speech and use the knowledge
to do something useful? There's a larger point, though. As I wrote last
month, too many major technology companies seem to be moving away from
R&D and innovation, instead spending huge sums on fighting the
patent wars or swallowing up little competitors at preposterous
valuations. Google spent $12. 5 billion to buy Motorola Mobility for its
patent portfolio, and Facebook forked over $1 billion to buy Instagram,
a company with no revenue, no business plan, and a cute little product
that's little more than a bell-and-whistle take on photo-sharing sites.
Old-fashioned innovation risk-taking at work
IBM, though, took a flyer on a product
that combines advances in artificial intelligence, big data, and natural
language processing to invent something new. When Big Blue started the
project, "we'd be hard-pressed to know if it could be monetized," says
Stephen Gold, who heads Watson Solutions. To me, that's what innovation
is all about: taking a risk and exploring the leading edges of
technology.
Gold did not care to say how much IBM
spent developing Watson. The project, which came out of the company lab
in Hawthorne, N.Y., took the services of 25 doctorate-level researchers
and four labs, so it wasn't cheap. But I'm willing to bet it cost a lot
less than the $1 billion shelled out for Instagram.
Did I drink a glass of Big Blue
Kool-Aid during my meeting with Gold at San Francisco's Palace Hotel
this week? I don't think so. If you look at the first real-world
applications of Watson, you'll notice they're aimed at solving
significant problems. Indeed, Watson may wind up saving lives -- and, oh
yes, make money for IBM and its shareholders.
Watson's $2.7 billion payoff
Suppose you're a doctor or a nurse
practitioner and a patient calls requesting an appointment, complaining
of a number of odd symptoms. When he comes in, you might have to order a
variety of expensive tests and do a fair amount of research to figure
out what's going on with him. Now suppose you could take the patient's
history and unstructured data from earlier appointments (notes written
on a tablet, for example), then comb through millions of articles from
medical journals and textbooks to get to the bottom of the issue. That's
exactly what WellPoint, an enormous health benefits company, is doing
in a pilot project using Watson.
The potential benefits are obvious:
quicker, more accurate diagnoses and fewer unnecessary expensive tests.
Starting soon is a partnership with Sloan-Kettering, a leading cancer
research institute, to study various forms of cancer with Watson's help,
first with lung cancer, then moving on to breast and prostate cancer.
Gold says that Watson is currently
"learning" about oncology and is being trained to answer questions about
it, in much the same way it was trained for the "Jeopardy" challenge.
An Associated Press report on the IBM/Sloan-Kettering partnership put it
this way: "Watson will be fed textbooks, medical journals and -- with
permission -- individual medical records. Then it will be tested with
more and more complicated cancer scenarios and assessed with the help of
an advisory panel."
As you may have noticed in the
"Jeopardy" events, Watson generally doesn't give definitive answers.
Instead, it calculates a probability that an answer is correct, an
important distinction in the world of medical research.
A third pilot program was announced in
March: Citigroup will use Watson to study ways to improve customer
service. Citigroup hopes that Watson's content analysis and
evidence-based learning abilities can improve customer transactions and
simplify the overall banking services. No, that's not curing cancer. But
landing Citigroup as a client is an important step in proving that
Watson can be a moneymaker.
How much money Watson can generate
still remains to be seen, of course. But in a research note late last
year, CLSA analyst Ed Maguire estimated that Watson may generate $2.7
billion in revenue in 2015, adding 52 cents of earnings per share. The
technology is so new that IBM is still looking at a variety of pricing
models. It will likely be related to results Watson achieves, Gold says.
IBM is not the juggernaut it once was,
nor is it quite the paradise for researchers it used to be. But a
company that can still spend some $6 billion a year on R&D and
embark on what Gold calls "a grand challenge" every decade is a
light-year ahead of companies pouring money down the rat holes of
useless litigation and frivolous acquisitions.
http://www.infoworld.com/d/the-industry-standard/beyond-jeopardy-how-ibm-will-make-billions-watson-192814
No comments:
Post a Comment