In 2019 Emma Haruka Iwao calculated pi to 31.4 trillion digits — a number so vast it took 170 terabytes to store, matching the Library of Congress.

On Pi Day 2019, Google's Emma Haruka Iwao set a world record by computing pi to 31,415,926,535,897 digits, a result so large it filled 170 terabytes — what Google compared to the entire print collection of the Library of Congress.

On March 14, 2019 — Pi Day — Google announced that one of its engineers, Emma Haruka Iwao, had calculated pi to 31,415,926,535,897 digits. That's roughly 31.4 trillion, and if you look closely at the number, you'll notice it isn't random: its leading digits spell out pi itself. The computation took so much storage that Google, in its own announcement on The Keyword (blog.google), described the 170 terabytes required as "roughly equivalent to the amount of data in the entire Library of Congress print collections."

That comparison is the part that sticks. The Library of Congress holds tens of millions of books, manuscripts, and documents. Iwao's calculation of a single number — admittedly a very long one — needed about the same amount of digital space to hold.

How Emma Haruka Iwao calculated 31.4 trillion digits of pi

Iwao didn't do this on a laptop. She ran the computation on Google Cloud, using 25 virtual machines working together over roughly 121 days. The software she used was y-cruncher, a program written by Alexander Yee specifically for computing constants like pi to absurd precision. The approach was the Chudnovsky algorithm, a formula developed by brothers David and Gregory Chudnovsky in the 1980s that converges on pi fast enough to make trillion-digit runs feasible.

The scale of the storage comes from the arithmetic itself. To multiply and verify numbers with trillions of digits, the machines constantly read and write enormous intermediate files to disk. That churn is what pushed the total to 170 terabytes — the figure reported at the time by The Seattle Times, which wrote that the process "required a staggering 170 terabytes, or about the same amount of data found in the print collections of the Library of Congress," and echoed by TechXplore in its own Pi Day coverage.

Guinness World Records verified the result as the most accurate value of pi ever calculated, beating the previous record of about 22.4 trillion digits set in 2016. For Iwao, the record had a personal arc: she has said she first became fascinated with pi as a child, and used calculating it as a benchmark when she was learning to program.

Why calculate pi to trillions of digits at all

The honest answer is that nobody needs 31.4 trillion digits of pi for engineering. NASA uses about 15 digits to navigate spacecraft across the solar system; with 40 digits you could calculate the circumference of the known universe to within the width of a hydrogen atom. The digits past that point have no practical use for measuring anything.

What they're good for is stress-testing. A run like this hammers the hardware for months, exposing weaknesses in processors, memory, storage, and the cloud infrastructure tying it all together. If a system can faithfully compute and verify trillions of digits without a single error, it can probably be trusted with far less exotic work. The record is partly a showcase for Google Cloud, and Iwao has been candid about that. It is also, plainly, a flex — a demonstration of how far the limits have moved.

Here's the seam worth arguing over. Mathematicians have long wondered whether pi is "normal" — whether its digits are distributed with perfect randomness, each digit and sequence appearing as often as any other over the infinite run. Decades of computed digits are consistent with normality, but consistency isn't proof. No one has ever proven pi is normal, and no finite calculation — not 31.4 trillion digits, not the quadrillions computed since — ever could. Each new record extends the evidence without settling the question.

So the number keeps growing, and the deepest thing about it stays stubbornly unproven.