The Economist Asia Edition - June 09, 2018

54 Business The EconomistJune 9th 2018

1

2

T

HE stereotype of a typicalMBAgradu-

ate is that of a confident, well-dressed

person who is destined for a career in

management consultancy, finance or

climbing the greasy pole at anS&P 500

company. The stereotype of an entrepre-

neur is a college drop-out. Yet business

schools, eager to prove that they are not

just factories for manicured professionals,

are increasingly keen to teach entrepre-

neurial skills to their students.

The temptation is to think that the abil-

ity and drive needed to start a business

cannot be taught. After all, who can en-

gender the combination of opportunism

and paranoia usually needed to start a

business? But some of those who have

taken the path argue that anMBAcourse

has several advantages. Shoshana Stew-

art, the chief executive of Turquoise

Mountain, a crafts business that started in

Afghanistan, who studied at the London

Business School (LBS), says anMBAgives

you three things; a network of people,

confidence and exposure, and an array of

skills.

The network effect can operate in sev-

eral ways. Oliver Samwer, who along

with his brothers founded the investment

group Rocket Internet in 2007, thinks the

guest speakers at theWHU-Otto Beish-

eim School of Management in Germany,

provided him with role models. “My

view is that it is all about the dream,” he

says. Every time a leader came to the

school, it inspired him to dream of a big-

ger, more global business. He has un-

doubtedly achieved lift-off: Rocket Inter-

net was valued at $8bn when it floated in

2014 and the Samwer brothers have in-

vested in several other successful technol-

ogy startups.

Sometimes the contacts are more im-

mediate. Vanessa Coleman started a busi-

ness called FINsix—which built an effi-

cient and compact power converter—at

MIT’s Sloan business school with three

other graduate students. They combined

their studies with their project, getting ini-

tial funding in the second year of the

course. And some of the advisers that

helped the company had connections to

MIT. A business school can also organise

events where budding entrepreneurs meet

potential investors and, in some cases,

those backers will be former students.

Self-belief is another quality that stu-

dents can gain fromthe classroom. Bilikiss

Adebiyi-Abiola came from Nigeria toMIT

and took a course run by Bill Aulet, a well-

known author, on entrepreneurship; she

says that helped her gain confidence in

pitching to a room full of investors.

When she went back to Africa she set

up a business which collects waste from

Nigerian households.The rubbish is sold

to recycling plants and the homeowners

get points, which canbe turned into cash.

Ms Stewart had worked for Turquoise

Mountain before taking herMBAatLBS.

While doing the course, she realised that

the business, which helps artisans with

marketing, sales and logistics, could ex-

pand into more countries; when she re-

turned, she expanded its operations to

Myanmar, Jordan and Saudi Arabia.

The final element is practical skills. Jon

Smith set up Pobble, a for-profit education

service, with his brother and others, just

before taking anMBAatLBS. Previously,

he had been a civil engineer and he says

that elements of the course were useful.

“Management accounting allowed me to

read a profit-and-loss account” he says,

while another course taught him how to

negotiate and bargain.

Mr Smith found that developing the

business while simultaneously studying

for the course was also helpful. “We saw

lots of case studies about what can go

right and wrong,” he says, adding that

“doing anMBAgives you the time to think

through what you care about.”

Clearly you do not need anMBAto

start a successful business. And plenty of

people takeMBAs and then continue on

to mundane corporate jobs. It is one thing

to start a business where the technical

skills learned in anMBAmaygofar.Itis

another thing to build that company into

a structure that can last for decades. Ms

Coleman, who has now left the business

she founded, says that the guidance pro-

vided by the school was less helpful in the

later stages.

But business schools will certainly

need to work harder to prove their rele-

vance; the cost of the qualification has

been rising and the number of applicants

has been falling. Around 11% fewer people

took theGMATtest (which acts as a de fac-

to entrance exam) in 2016 than in 2012. If

business schools can improve the skills of

those who try to build companies, that

has to be good news. We have enough

management consultants already.

Bartleby Start them up

Business schools can give entrepreneurs vital skills

Economist.com/blogs/bartleby

been to design processors from the ground

up with AIin mind. The result of Graph-

core’s efforts is called an intelligent pro-

cessing unit (IPU). This name is not just

marketing: on GPUs, memory (the staging

area for data) and brain (where they are

processed) are kept separate—meaning

that data constantly have to be ferried back

and forth between the two areas, creating a

bottleneck with data-heavyAIapplica-

tions. To do away with it, Graphcore’s

chips do not just have hundreds of mini-

brains, but the memory is placed right next

to it, minimising data traffic.

Graphcore’s chip can also hold entire

neural networks, computational models

inspired by structures in biological brains,

which are used in manyAIapplications.

Having such models, which can be im-

mensely complex with billions of parame-

ters, sit in the chip allows them to be

“trained” more quickly—the act of feeding

them with lots of data (pictures of cats,

say), so they learn to recognise them. The

set-up also simplifies what is known as “in-

ference”, when the model applies what it

has learned (spotting cats, for instance).

Cerebras is going further still. It is not

only designing a new processor, which is

similar to Graphcore’s, but a specialised AI

computer as well. Puttinga new chip on a

circuit board, as Graphcore does, that is

added into an existing system limits spe-

cialisation and optimisation because of

constraints in power, cooling and commu-

nication, says Mr Feldman. But this means

that he has a steeper hill to climb: while

Graphcore has already delivered a first

batch to customers, Cerebras has yet to an-

nounce when its product will be available.

Although Graphcore and Cerebras

were early to see the need for specialised

AIchips, they are by no means alone. Doz-

ens of startups are creating what are

known as “application-specific integrated