*
New structure makes OpenAI 'IPO-able'
*
Says 10 gigawatts of data capacity could cost $500 bln
*
Points to rapidly growing search market as priority
(Adds quote on possible $500 billion data centre investment,
search as a priority in paragraphs 11-14)
By Conor Humphries
DUBLIN, May 28 (Reuters) - OpenAI's restructuring plans
position it for a potential future IPO, but any such decision
would depend on the mood in public markets as well as the
readiness of the company, Chief Financial Officer Sarah Friar
said on Wednesday.
OpenAI, in which Microsoft ( MSFT ) has invested more than
$13 billion, outlined plans in December to convert its
for-profit arm into a public benefit corporation (PBC), a
structure designed to balance shareholder returns with social
goals, unlike nonprofits, which are solely focused on public
good.
The ChatGPT-maker dialed back the plan earlier this month,
giving the nonprofit parent control of the PBC through a big
shareholding, while still allowing the for-profit arm to raise
more capital to keep pace in the AI race.
"A PBC gets us to an IPO-able event ... if and when we want
to," Friar told the Dublin Tech Summit.
"Nobody tweet in this room that Sarah Friar just said
anything about OpenAI ultimately going public," she added. "I
did not. I said it could happen."
Asked what would influence a decision, Friar said that as
with any company planning to launch on the stock market, the
company and markets would have to be ready.
"You can show up at the altar all ready to go, and if the
market's not ready for you, yeah, you're just out of luck," she
said.
"Which is why you have to build a company that can be
sustainable and safe regardless of where the public markets are,
how open that window is."
To be a public company, "you definitely need some sense of
predictability," Friar added.
"The market will put up with a certain degree of
unpredictability. Particularly when growth is high ... but the
market doesn't really love it."
To give a sense of the "massive" scale of capital that
OpenAI might require, Friar said that while a 1 gigawatt data
centre footprint costs about $50 billion, the company's
"appetite and ambition" over the next couple of years was to get
closer to about 10 gigawatts.
Friar singled out the rapidly growing AI search market as a
priority and said that while the company would seek
efficiencies, its focus was on finding the next breakthrough
product.
"The search market is becoming a big market," Friar said.
"In that world, I don't really want people spending an
inordinate amount of time trying to save an extra 1% when I
would rather they went out and kind of built the next
state-of-the-art product."
(Writing by Conor Humphries
Editing by Mark Potter and Elaine Hardcastle)