|
|
AI requires a significant amount of computation to train models and run inferences. Over the past decade, as models have become increasingly complex, the computational demand has grown exponentially. For example, OpenAI found that from 2012 to 2018, the computational demand for its models doubled every two years to doubling every three and a half months. This has led to a surge in demand for GPUs, with some cryptocurrency miners even repurposing their GPUs to provide cloud computing services. With intensifying competition for access to computation and rising costs, some projects are leveraging blockchain technology to offer decentralized computing solutions. They provide on-demand computation at competitive prices, allowing teams to train and run models affordably. In some cases, there's a trade-off between performance and security.
The demand for state-of-the-art GPUs, such as those produced by Nvidia, is enormous. In September 2023, Tether acquired shares in the German Bitcoin miner Northern Data, reportedly spending $420 million to purchase 10,000 H100 GPUs (among the most advanced GPUs for AI training). The wait time for obtaining top-notch hardware can be at least six months, often even longer in many cases. Furthermore, companies are frequently required to sign long-term contracts to obtain computing power they may not even use. This can result in situations where there is compute available but not accessible in the market. Decentralized computing systems help address these inefficiencies in the market, creating a secondary market where compute owners can immediately sublet their excess capacity upon receiving notice, thereby releasing new supply.
In addition to competitive pricing and accessibility, a key value proposition of decentralized computing is censorship resistance. Cutting-edge AI development is increasingly dominated by large tech companies with unparalleled computing and data access capabilities. The first key theme highlighted in the AI Index Report 2023 Annual Report is that the industry is increasingly surpassing academia in the development of AI models, consolidating control in the hands of a few technological leaders. This raises concerns about whether they have the ability to exert significant influence in shaping the norms and values that underpin AI models, especially after these tech companies push for regulations to limit the development of artificial intelligence beyond their control. |
|
Post time 25-2-2024 20:43:46