What a difference half a year can make in the cryptocurrency world. Crypto mines have been all but abandoned following the fall of Bitcoin and its contemporaries — at least compared to the gold rush of late 2017. And the global juggernaut of GPU processing power assembled in a mad rush to cash in on the boom now lays significantly dormant. Smart investors and financial experts are waiting with bated breath to see where that juggernaut will focus its attention next.
It has become apparent that there are only a handful of fields with the potential to satiate the beast: AI computing, predictive analytics, and video rendering. AI, in particular, is quickly becoming an integral part of every major industry, including medicine, finance, telecommunications, insurance, design, healthcare, defense, and transportation. Annual data center revenue has already surpassed US$50 billion. At the pace AI is developing, all of that under-utilized processing power amassed to mine cryptocurrency could soon be a hot commodity.
This will be the next gold rush for former miners — and rather than digging for digital gold, they will make their money powering breakthroughs in science and the arts. The trick is finding a way to connect miners with companies that have this growing demand for AI computing. Tatau has the solution to make that transition possible.
The Tatau Platform makes the process painless and intuitive. Tatau takes inputs from buyers (or customers), breaks the work up and sends it to its network of customer (such as crypto miners), manages all the tricky bits like setting up escrow contracts and ensuring the work is done correctly, and then makes sure the buyer only pays for computing completed correctly. At the core is the Tatau token. The tokens can be used as both a secure form of payment when buying or selling computational output, and as a method of staking computational nodes and validators.
So the juggernaut is poised, and Tatau has the key, but who will drive the demand?
Michael Werth uses high-performance computing in his work with convolutional neural networks, AI and computational physics in the aerospace and defense industry. He uses a variety of resources to process the data he needs.
“Sometimes I use DGX Stations, sometimes more traditional supercomputers with GPU compute nodes — with a variety of different Tesla generations — and sometimes I re-use my gaming box if I’m dabbling at home.”
Michael says a solution like the one offered by Tatau — matching up buyers and suppliers of GPU processing power — would be a very lucrative one.
“That would be of interest to me and others in my field. It would be a huge boon to countless academics and hobbyists, including professionals sharpening their skills at home, to have more affordable access to computing resources.”
“I have a GTX 1080 Ti in my desktop for my personal projects, but it would be great to have access to more resources that simply bill me for time spent. I’ll always prefer using the Nvidia Tesla V100 to a GTX 1080 Ti, but the V100 is more than 10 times more expensive to purchase and a lot less commonly available.”
Michael believes that the lower costs of Tatau’s solution would make it very attractive to many people like him and companies like Boeing, as well as being a much more worthwhile option for ex-miners.
“People who are still running dedicated GPU mining rigs would almost definitely be happy to have a well-paying alternative now that profitability has decreased significantly for GPU mining. And casual former-miners — who significantly outnumber the dedicated miners — would also be happy for a new well-paying outlet. What’s key here is simply offering a better deal than what cryptocurrency mining offers and making it very easy for users to make their hardware available.”
“The mathematics of home-mining rigs never added up unless you could somehow obtain free electricity, or unless you assumed cryptocurrencies would continue increasing in value exponentially forever.”
Michael believes the key is solving the ‘honesty problem’ of validating that the AI computations have been completed — and done correctly. Tatau does this by using machine learning techniques to scrutinize results, therefore policing the output of correct computations.
“Solving the honesty problem is a pretty big game changer, provided it is coupled with an intuitive user interface. People with hardware want something like NiceHash, where you just click a button and slowly get paid. On the other side, submitting jobs to that kind of network also needs to be intuitive. I can’t emphasize the need for an intuitive user interface enough. If someone can offer that, then you’ll find a lot of people who want to offer their hardware.”
Another lucrative application for all that processing power is 3D video rendering. Darkroom is a New Zealand design agency that creates innovative event visuals, videos, and installations for festivals such as Glastonbury and companies like Louis Vuitton. They produce an immersive experience called The Trip for Les Mills gyms — basically, a roller-coaster ride of trippy visuals projected onto all four walls of a cycle class — and that type of 360-degree video requires a lot of computational power to render. But since they don’t need to render that type of file every day, it’s not practical for companies like Darkroom to invest in a wall of GPUs.
“At the moment we have 24 cards — a mix of GTX980Ti, 1080 and 1080Ti cards — which gets us by most of the time,” says Darkroom owner and creative director Bruce Ferguson.
“But about once every three months we need some serious 3D graphics rendering, using Octane Render. That takes about two weeks’ worth of time using eight 1080Tis.”
At the moment the process is a bit finicky to set up.
“We usually rent machines from UltraRender.com and we use team viewer or something similar to remote in. We need to set up various software on each machine — like Maxon C4D, Octane Render, and Dropbox, plus plugins — then we use the machines as we would if they were local, using Dropbox to sync the results to our local drives.
Ferguson says Tatau’s solution could make the process much simpler, by matching up their requirements with machines that were ready to use without any extra setup.
“If there was a community of C4D octane render machines on the blockchain that would be great.”
Ferguson says that as an added bonus they could also use Tatau to lease the Darkroom rendering power back to similar companies. This would give them an additional source of income. It can be hard to justify buying expensive graphics cards and software licenses if you don’t use them to their full potential every day, so Tatau’s solution would give them a better return on their initial investment.
A Question of Sustainability
Now that it’s clear Bitcoin’s meteoric rise in value was not sustainable in the long-term, questions are also being asked about the sustainability of its electricity consumption. The machines doing the mining are consuming astronomical amounts of energy — the Digiconomist Bitcoin Energy Consumption Index estimates miners use over 70TWh per year. That’s enough to power 6,750,000 US households or more than the entire annual energy consumption of Austria.
Not only will Tatau help transition hardware away from unsustainable mining, but it will also help transition larger companies from CPU to GPU processing, which is better for the environment and their bottom line. At the moment, many companies that require high-performance computing pay for the use of major CPU server farms provided by the likes of Amazon. But the growth in fields such as AI and video rendering has left CPUs struggling to keep up. Luckily, GPUs can process data at greater speeds than CPUs using a massively parallel approach. So by moving to distributed GPU processing, as offered by Tatau, companies can save time, money and even reduce their carbon footprint.
“In my mind, Tatau is the Uber of compute”, says Tatau Co-Founder and CEO, Andrew Fraser. “When Uber goes into a new market, they don’t go out there and buy a whole heap of taxis. Instead, they turn every car on the road into a potential taxi. We’re doing the exact same thing for the massive global supply of under-utilized GPUs. We can be the largest supplier of compute in the world without owning any computers.”
The sooner Tatau can redeploy the crypto mines, the better for business, AI innovation and future generations.