London-based Magic Pony went from A.I. startup to a multimillion dollar cash-out in 18 months. Was selling to Twitter the right exit strategy?
Since 2010, the biggest tech companies had been on an acquisition spree for Artificial Intelligence (A.I.) startups. As of August 2019, there had been 635 acquisitions since 2010, with Apple leading the tech behemoths with 20 including the 2017 deal for RealFace, followed by Google with 14. Microsoft (10), Facebook (eight), and Amazon (seven) were all active.
In June 2016, Twitter bought London startup Magic Pony for a reported US$150 million. Started by Rob Bishop and Zehan Wang just 18 months prior, Magic Pony had developed technology to deliver high quality video streaming over low internet bandwidth. Twitter, which had acquired Periscope in 2015 and the now-discontinued Vine in 2012, was focusing on its video capabilities and decided Magic Pony would be an ideal acquisition.
A multi-million dollar exit in 18 months is a good outcome for any startup, but was it the right time to sell? And was Twitter the right company to sell it to?
clear (A.I.-ENHANCED) vision
Bishop and Wang co-founded Magic Pony in 2014 after graduating from Imperial College London; Bishop had a Masters degree in signal processing while Wang was a PhD in visual image processing. Although they attended the same school, the co-founders did not know each other until they met at Entrepreneurs First, an entrepreneurship and startup incubator.
A.I. startups in the UK were popular acquisition targets for Silicon Valley heavyweights in the mid-2010s. Google bought DeepMind for US$582 million in 2014, followed by Apple’s bid for speech recognition startup VocalIQ in 2015 and Microsoft the following year with a US$250 million offer for Swiftkey, a predictive keyboard for smartphones.
The pace of A.I. startup acquisitions was picking up, and Bishop and Wang appear to the right combination of skills to address a key issue: low-quality image quality despite ever-increasing computing power. The technology they would employ would be Deep Learning, a kind of unsupervised machine learning without explicit human instructions that is best illustrated by Google’s 2012 simulation demonstration that recognised cats from millions of random videos without being told what a cat looked like. Andrew Ng, the Stanford professor and Google research partner who developed the simulation, describes it as follows:
“The point wasn't to find a cat. The point was to have a software, maybe a little simulated baby brain, if you will, wake up, not knowing anything, watch YouTube video for several days and figure out what it learns. And I'm sure it's learned tons of other things other than, you know, cats. And it's just that cats was one thing we happen to look for and found.”
The co-founders decided to optimise quality technology that delivered high quality streaming over low Internet bandwidth, something that no other startup appeared to be doing. The team had created techniques that worked on low-resolution, compressed videos taken using smartphones and shared on social media. Its technology could run on a standard graphics card installed on desktop computers and smartphones, therefore making it accessible to most devices that people were using.
Show me the money!
Bishop and Wang secured seed funding of US$16,000 from Entrepreneurs First and then followed that up with another US$6.3 million by 2016 from two London-based backers, Octopus Ventures and Balderton Capital. Their early technology demonstrations had so amazed an early investor who described it as a “magic pony”, which the co-founders adopted as the company name.
By now, Magic Pony was starting to gain a bit of attention. Whenever Bishop and Wang pitched their product, potential customers instead made offers to acquire Magic Pony. In May 2019, 18 months after the founding of the company, Magic Pony was valued at an estimated US$30 million and held about 20 patents.
When Twitter made its reported US$150 million offer for Magic Pony, it appeared they would be getting technology tailor-made for the Periscope app whose performance was often limited by mobile network connectivity, bandwidth, camera phone resolution as well as the lighting conditions under which the video was broadcast. Magic Pony bridged that gap by automatically improving video quality and therefore enhancing overall viewing experience.
With the acquisition, Twitter would be keeping up with its biggest competitors Google, Facebook and Microsoft in the A.I. race. Besides Magic Pony’s patents, Twitter will also get 11 machine learning and computer vision PhDs who were kept with retention bonuses. The team would join Twitter’s machine learning division while Bishop would move to the Twitter HQ in San Francisco.
It appears to be a perfect ending: Twitter get what they need to keep up, and two entrepreneurs realise the big cash-out all startups dream about. But is it?
This is an adapted version of the SMU Case, "Twitter's acquisition of Magic Pony technology in the race for AI". To see the full case, please click on the following link: https://cmp.smu.edu.sg/case/4186
Follow us on Twitter (@sgsmuperspectiv) or like us on Facebook (https://www.facebook.com/PerspectivesAtSMU)
Last updated on 28 Nov 2019 .
