BloombergBusinessweek|SoonerThanYouThink June 10, 2019
S
imon Knowles, chief technology officer of
GraphcoreLtd.,issmilingata whiteboardashe
mapsouthisvisionforthefutureofmachinelearn-
ing.Heusesa blackmarkertodotanddiagramthe
nodesofthehumanbrain:thepartsthatare“rumina-
tive,thatthinkdeeply,thatponder.”Hisstartupis try-
ingtoapproximatetheseneuronsandsynapsesinits
next-generationcomputerprocessors,whichthecom-
panyis bettingcan“mechanizeintelligence.”
Artificialintelligenceis oftenthoughtofascomplex
softwarethatminesvastdatasets,butKnowlesandhis
co-founder,ChiefExecutiveOfficerNigelToon,argue
thatmoreimportantobstaclesstillexistinthecomput-
ersthatrunthesoftware.Theproblem,theysay,sit-
tingin theirairyofficesin theBritishportcityofBristol,
isthatchips—known,dependingontheirfunction,as
CPUs(centralprocessingunits)orGPUs(graphics
processingunits)—weren’tdesignedto“ponder”in
anyrecognizablyhumanway.Whereashumanbrains
useintuitiontosimplifyproblemssuchasidentifying
anapproachingfriend,a computermighttrytoana-
lyzeeverypixelofthatperson’sface,comparingit toa
databaseofbillionsofimagesbeforeattemptingtosay
hello.Thatprecision,whichmadesensewhencomput-
erswereprimarilycalculators,is massivelyinefficient
forAI,burninghugequantitiesofenergytoprocessall
therelevantdata.
WhenKnowlesandthemorebusiness-mindedToon
foundedGraphcorein2016,theyput“lessprecise”
computingattheheartoftheirchips,whichtheycall
intelligenceprocessingunits,orIPUs.“Theconceptsin
yourbrainarequitevague.It’sreallytheaggregationof
veryapproximatedatapointsthatcausesyoutohave
precisethoughts,”saysKnowles,whoseEnglishaccent
andfrequentchuckleinvitecomparisonstoa Hogwarts
headmaster. (Given his constant whiteboard pontif-
icating, Toon jokingly addresses him as “Professor
Knowles.”) There are various theories on why human
intelligence forms this way, but for machine learning
systems, which need to process huge and amorphous
information structures known as “graphs,” building a
chip that specializes in connecting nodelike data points
may prove key in the evolution of AI. “We wanted to
build a very high-performance computer that manipu-
lates numbers very imprecisely,” Knowles says.
Put another way, Graphcore is developing a brain
for computers that, if its co-founders are right, will
be able to process information more like a human
instead of faking it through massive feats of number
crunching. “For decades, we’ve been telling machines
what to do, step by step, but we’re not doing that any-
more,” Toon says, describing how Graphcore’s chips
instead teach machines how to learn. “This is like
going back to the 1970s—we need to break out our
widelapels—whenmicroprocessorswerefirstcoming
out.We’rereinventingIntel.”
Investor Hermann Hauser, co-founder of Arm
HoldingsPlc, which controls the most widely used chip
designs, is betting that Knowles and Toon’s IPUs will
unleash the next wave of computing. “This has only
happened three times in the history of computers,”
Hauser says—CPUs in the 1970s, GPUs in the 1990s.
“Graphcore is the third. Their chip is one of the great
new architectures of the world.”G
raphcore’s origins lie in a series of sympo-
siums Hauser organized in 2011 and 2012 at the
University of Cambridge for the Royal Society, the
scientific fellowship that counts Isaac Newton and
Charles Darwin as alums. Around a posh dining room
at King’s College, AI experts, neuroscientists, statis-
ticians, and zoologists debated the impact advanced
computing would have on society.
Knowles, whom Hauser says “has a brain the size
of a globe,” felt out of place in this ivory tower, even
thoughhegothisstartatCambridge.Aftergraduat-
inginthe1980s,hestudiedearlyneuralnetworksata
U.K.governmentresearchlab.Hewentontoco-create
Element14,a wirelessprocessorstartupthatsoldto
BroadcomInc.in 2000 for$640million. Soon after, he
and Toon, who also had experience building semicon-
ductorstartups,teamedupforthefirsttime.In 2002
theycreatedIcera,a mobilechipmaker,whichthey
soldtoNvidiaCorp.for$436million a little less than a
decade later. (Nvidia has since shuttered it.) The two
weren’t ready to retire at that point. “We’re both crap
at golf,” Toon says.
They were batting around other ideas when Knowles
went to the Cambridge lecture series. “I was the
scruffy chap in the room with a stovepipe hat who just
wanted to build stuff—you know, ‘Never mind thermo-
dynamics,I wanttobuilda steamengine!’” herecalls.
WhenSteveYoung,a Cambridgeprofessorofinforma-
tionengineeringwholatersolda speech-processing
service to Apple Inc. now used in Siri, gave a presen-
tation on the limits of computational dialogue systems,
Knowles peppered him with questions about energy
efficiency. “I asked what numerical precision he was
using for his arithmetic, which to Steve seemed out of
left field,” says Knowles, who stresses that “the pre-
cision of numbers are very critical as determinants of
energy” in silicon.
Dayslater,YoungemailedKnowlestosayhisstu-
dentsinvestigatedthematteranddiscoveredthey
wereusing 64 bits of data per calculation. They real-
izedtheycouldperformthesamefunction,asKnowles
hadsuggested,withlessprecisearithmetic,using
8 bits. When the computer had less math to do,61