Since generative AI burst into global consciousness in 2023, unprecedented demand for computing power has emerged along with demand for applications that use this technology.
The tool is like OpenAI's ChatGPT They require thousands of Nvidia GPUs (graphics processing units) to smoothly process all the information coming in and out. Last week, Nvidia compared GPUs to rare earth metals for AI, saying they are "fundamental" to how today's generative AI works.
The energy needed to power all this hardware is the equivalent of that of a small country, according to a Report published by the French energy company Schneider Electric last year. On Wednesday, OpenAI CEO Sam Altman told an audience in Davos that an energy breakthrough was needed to drive AI advances. "There's no way to get there without a breakthrough," he said, suggesting it was a motivation to invest more in nuclear fusion.
Fortune Business Insights estimated earlier this year that the global GPU market size was valued at $2.39 billion in 2022 and is projected to grow from $3.16 billion in 2023 to $25.53 billion in 2030. Nvidia claims that more than 40,000 Companies use Nvidia GPUs for artificial intelligence and accelerated computing. .
To meet the demand, Nvidia announced in August it would be tripling its GPU production. In November, Microsoft signed a multi-year agreement with Oracle to supply computing power for its Bing Chat AI functionality.
Now companies that once served the cryptocurrency mining boom are pivoting to take advantage of the latest data gold rush.
Canadian company Hive Blockchain changed its name in July to Hive Digital Technologies and announced it was shifting toward AI.
โHive has been a pioneering force in the cryptocurrency mining sector since 2017. The adoption of a new name signals a significant strategic shift to harness the potential of GPU Cloud computing technology, a vital tool in the world of AI, machine learning and advanced technologies. data analytics, allowing us to expand our revenue channels with our fleet of Nvidia GPUs,โ the company said in its announcement at the time.
The company's chief executive, Frank Holmes, told Guardian Australia the transition required a lot of work.
โMoving from mining Ethereum to hosting GPU cloud services involves purchasing powerful new servers for our GPUs, upgrading network equipment, and moving to higher-tier data centers,โ he said.
โThe only thing in common is that GPUs are the workhorses in both cases. GPU cloud requires high-end supporting hardware and a more secure and faster data center environment. โThere is a steep learning curve in the GPU cloud business, but our team is adapting well and learning quickly.โ
For others, like Iris Energy, a data center company operating in Canada and Texas and co-founded by Australian Daniel Roberts, that has been the plan all along. Iris did not require any changes to the way the company operated when the AI โโboom hit, Roberts told Guardian Australia. saying.
โOur strategy has really been to start the data center platform with bitcoin mining and then just preserve optionality throughout the digital world. The difference between us and crypto miners is that we are not actually miners, we are data center people.โ
The company still touts its bitcoin mining capabilities, but in the latest results, Iris said it is well positioned for "power-dense computing" with 100% renewable energy. Roberts said it was not a situation between bitcoin mining and AI.
โI think when you look at bitcoin versus AI, the market will simply reach equilibrium based on the market demands for each product,โ he said.
โSo bitcoin is in demand as a store of value, gold 2.0... will rise, it will generate an economic incentive to secure it.
โOn the contrary, as AI adoption grows there, people will be willing to pay for it. And then for us, we have the option to switch between the two and optimize based on the route.โ
Holmes said Hive also saw the two industries operating in parallel.
โWe love the bitcoin mining business, but its income is quite unpredictable. GPU cloud services should complement it well,โ he said.
"Revenues should be more stable, but still offer attractive margins and the opportunity for rapid growth."
Despite what had been declared a โcrypto winterโ in 2023, the value of bitcoin hit a two-year high of $49,061 on January 11, after the US securities regulator. approved first US-listed exchange-traded funds (ETFs) to track bitcoin. This week it fell below $40,000 again.
As with cryptocurrency mining, the enormous computing power required by AI systems means huge amounts of energy and carbon emissions for some of the centers.
AI companies like OpenAI keep their carbon emissions figures secret, but has been estimated that training the previous version of GPT, GPT-3, consumed 1,287 megawatt hours of electricity and generated 552 tons of CO2, the equivalent of 123 fossil fuel-powered cars driven for a year.
Iris Energy believes that the use of renewable energy is not only better for the environment, but also a cost saving.
"We've gone to the source of surplus low-cost renewable energy where we've gotten a lot of land and a lot of energy," Roberts said. โ[At our] In Texas, we have a 600 megawatt grid connection in an area where there are 32 gigawatts of wind and solar energy. And the transmission line is 12 gigawatts to export to Dallas and Houston.โ
While there is a lot of buzz, some view the shift in the market with some skepticism, suggesting that some may be jumping from one fad to the next.
Institutional Investor reported in August that a โpenny stockโ company known as Applied Sciences had reinvented itself as a bitcoin miner hosting company in April 2022 as Applied Bitcoin, but in November 2022, perhaps sensing the shift in investment, rebranded as Applied Digital and focused on AI.
