- Facebook announced that it will open-source Big Sur, its latest hardware designed for AI computing at a large scale.
- Alphabet recently open-sourced its second-generation Artificial Intelligence (AI) system TensorFlow.
- Why are top tech giants open-sourcing sophisticated and very advanced AI algorithms, development of which costs a lot of money?
In November, Alphabet Inc-A (NASDAQ:GOOGL) announced that it was open-sourcing its second-generation Artificial Intelligence (AI) system TensorFlow. TensorFlow, originally developed for advanced machine learning and deep neural networks research, is now a highly general and scalable machine learning system used in many Alphabet research projects and production lines.
Facebook (NASDAQ:FB), which is becoming one of the most advanced AI research centers in the world, claiming that its AI developments have accelerated to the point that AI will surely shape the way humans interact with computers, has recently announced that it will open-source Big Sur, its latest hardware designed for AI computing at a large scale.
Big Sur incorporates eight high-performance GPUs of up to 300 watts each, with the flexibility to configure between multiple PCI-e topologies. Compatible with Open Rack, the first rack standard designed for data centers, Big Sur is twice as fast as Facebook's previous generation hardware, which means it can be trained twice as fast and explore networks twice as large. And distributing training across eight graphic processors (GPUs) allows the size and speed of Facebook's neural networks to be scaled by another factor of two.
Using GPUs to accelerate AI software permits important advances in speech, image, and language understanding. The Big Sur hardware can be used to run Google’s TensorFlow software.Facebook worked closely with NVIDIA (NASDAQ:NVDA), a leading manufacturer of GPUs, on its new Big Sur server technology, which leverages NVIDIA's Tesla Accelerated Computing Platform. It's worth noting that NVIDIA itself wants to be a player in the AI sector and recently unveiled a credit card -sized embedded computer designed for AI applications.
Open design materials for Big Sur will be submitted to the Open Compute Project (OCP), spearheaded by the social networking giant.
"We want to make it a lot easier for AI researchers to share techniques and technologies," states the Facebook announcement. "As with all hardware systems that are released into the open, it's our hope that others will be able to work with us to improve it. We believe that this open collaboration helps foster innovation for future designs, putting us all one step closer to building complex AI systems that bring this kind of innovation to our users and, ultimately, help us build a more open and connected world."
Previously in February, Facebook announced that it was open-sourcing optimized software modules for Torch, an open-source development environment for deep machine learning and computer vision used at a number of academic labs as well as at Google, NVIDIA, and Intel (NASDAQ:INTC).
IBM (NYSE:IBM), another top tech company at the leading edge of AI research with its Watson AI platform and next-generation "neuromorphic" (brain-like) computing chips, recently transferred its SystemML machine-learning software system to the Apache Software Foundation, home of several major open-source projects. Derek Schoettle, general manager of IBM Cloud Data Services unit, said that machine-learning technology has to be opened up for it to become widespread.
In an article titled "Facebook Joins Stampede of Tech Giants Giving Away Artificial Intelligence Technology," MIT Technology Review notes a trend - top tech giants are open-sourcing sophisticated and very advanced AI algorithms whose development costed them a lot of money - and wonders what it means and why these companies are giving precious intellectual property away.
"Companies like us actually thrive on fast progress; the faster the progress can be made, the better it is for us," said FAIR Director Yann LeCun. High profile commitment to open-source software can also boost the image of the company and make it easier to attract and keep top talent. But LeCun added that one reason - plausibly the main reason - to open-source Big Sur and key proprietary algorithms and software is that Facebook is well positioned to exploit any resulting improvements and breakthrough, faster than other companies.
In fact, AI is a computing intensive technology - the AI algorithms of Alphabet, Facebook, and IBM are meant to run on supercomputers and large data centers. Individual developers and small companies can help innovating on software and even hardware, but only the tech giants have enough cash to build huge data centers with hardware optimized for AI applications. Therefore, it seems likely that, by open-sourcing their technology, Alphabet and Facebook will be able to benefit from crowd-sourced innovation at little risk. This move confirms Alphabet and Facebook's stock as a strong buy-and-hold.