Google at 25: By the numbers

 

When Google launched in 1998, our ambition was to create a search engine that would help people make sense of the information they find online — and that’s still true today. But over the past 25 years, Google has become more than a search engine: Today we map the world, bring AI into everyday tools and help people gain skills in digital careers, among other things.

For our 25th birthday, we’re taking a look at Google by the numbers (25 of ‘em, to be exact) to reflect on our origins and see how far we’ve come.

A googol

A one with 100 zeroes — and also, the inspiration for our name.0The 100-trillionth digit of pi, which a Google researcher discovered in 2022.

Go deeper: Thanks to advances in cloud computing, in 2022 a Google researcher broke her own record of reaching new digits of pi. To do this, Emma Haruka Iwao used a computer program run on Google Cloud to process this data as quickly as possible — and it found the 100-trillionth digit at a new, speedy record of 157 days. To put 100 trillion digits in perspective, just know that 100 trillion inches of pie crust can stretch from the earth to the moon and back roughly 3,304 times.70 billion

The average number of daily views on YouTube Shorts, as of this month.

250,000

How many patients globally our ARDA toolhas screened for diabetic retinopathy, a leading cause of preventable blindness.

Go deeper: Diabetes is a growing issue globally, and because there aren’t enough vision experts in certain parts of the world, related complications like diabetic retinopathy can go undiagnosed and untreated. Together with Verily, we set out to see if we could train an AI model that could identify diabetic retinopathy. The result was ARDA, or Automated Retinal Disease Assessment, which has been found to perform on par with eyecare doctors in screening for the condition.70%

How many generative AI unicorns are GoogleCloud customers.

Go deeper: A “unicorn” in this sense is a startup with a value of over $1 billion.

Unicorns

12 billion

The number of visual searches conducted using Google Lens every month.

5,000+

How many Google Doodles we’ve created.

Go deeper: In addition to the more than 5,000 Doodles we’ve created in-house at Google, the team receives around 7,000 submissions from Googlers as well as users (includingDoodle forpublished and learn more about them and the artists who created them.

8.5 million

The concurrent viewers of India's moon landing on the Indian Space Research Organization’s main YouTube channel — the most-viewed livestream on YouTube ever.

Go deeper: On August 23, 2023, India’s Chandrayaan-3 spacecraft landed on the south pole of the moon, making India the fourth country to land on the moon and the first to land on its south pole. While on the moon, the Chandrayaan-3 will collect and analyze data about the lunar environment.

30

The maximum number of seconds you have to unsend an email in Gmail.

Go deeper: Accidentally sent an email with a glaring typo in it? Just press unsend! You can customize your unsend email window by heading to settings — unsend time lengths include 5, 10, 20 and 30 seconds.10 million+

The number of miles we’ve mapped with the Street View car — enough to circle the world more than 400 times.

Go deeper: We’ve also used tricycles, sheep, snowmobiles and camels to capture Street View imagery around the world.,10 million

The number of cells a Google Sheet can contain (in case you're wondering, that's 18,278 columns).

20+

How many coding languages Bard knows.

Go deeper: That list includes Python, JavaScript, C++, Java, R, SQL, C, Go, Swift and Kotlin — but Bard is always learning new languages, so this number will grow!133

The number of languages Google Translate supports.

Go deeper: Advances in AI and machine learning have helped make this possible. Some languages are "under-resourced," meaning it's hard to find high-quality datasets to train our systems to accurately offer translations for these languages. In 2022, we had a breakthrough: For the first time, we added 24 languages to Translate using Zero-Shot Machine Translation, which is where a machine learning model learns how to translate something with no language-specific translation examples. And while AI was a big part of this achievement, our partnership with native speakers to create parameters and verify data was also crucial; when they evaluated the results, they found that the majority of the translations were high quality.