The recent release of DeepSeek has disrupted the disrupters. However the conclusions and “buzz” about what it all means have logical weak points, and some apparent “wrong headedness” involved.
One of the most important factors of DeepSeek’s announcements is their claim to greatly increased efficiencies using innovative new algorithms and software on less capable and less expensive GPUs. Greater efficiencies DO reduce costs.
However as part of an OpenSource release, those software efficiencies are now available to the larger frontier model companies as well, and not just moms and pops.
More importantly, there’s something called Jevon’s Paradox that seemingly applies, meaning that greater efficiencies drive greater use, and thus increased opportunities for profit. If one uses that perspective then we are still in the world where the number and capability of GPUs is critically important, as is the size and number of data centers, and the amount of electric production capability needed to drive them.
IOW, despite DeepSeek, the size equations favoring giant tech players are still relevant, if perhaps a bit less determinative, depending on the business plans involved.
From ChatGPT advanced voice conversation:
Jevons’ Paradox
Definition:
Jevons’ Paradox states that as technological improvements increase the efficiency of resource use, the overall consumption of that resource may actually increase rather than decrease. This occurs because increased efficiency reduces costs, making the resource more accessible and leading to greater demand.Origin:
The paradox was first observed by economist William Stanley Jevons in the 19th century while studying coal consumption in England. He noted that as steam engine efficiency improved, coal consumption increased rather than decreased, as industries found more ways to use the now more cost-effective energy source.Modern Applications:
Jevons’ Paradox applies broadly across different fields, including:Energy Efficiency: Improvements in fuel efficiency can lead to greater overall energy consumption due to increased travel and industrial expansion.
Computing Power: As computers become faster and cheaper, we use them for more applications, increasing overall demand for processing power and data storage.
AI & Automation: Cost reductions in AI-driven tasks make AI more widely adopted, expanding overall AI usage rather than limiting it.
Implications:
Jevons’ Paradox suggests that increasing efficiency alone may not be enough to reduce total resource consumption.
Policy measures such as carbon taxes, regulatory caps, or alternative incentives may be needed to ensure that efficiency gains translate into actual reductions in resource use.
Conclusion:
Jevons’ Paradox is a critical concept in economics and sustainability discussions, illustrating that greater efficiency does not always lead to lower consumption. Instead, market forces and human behavior often drive increased usage, requiring additional policies to manage overall resource demand.