Artificial intelligence revolution – what is the real cost of telecommunications?

The stratospheric growth of AI performance incorporated into applications has led many experts to reveal serious concerns about the readiness of telecommunications companies to handle rapidly increasing demands. Content-generating tools like ChatGPT are seeing a huge increase in computing power, which is likely to have a huge impact on network capacity, energy and the environment.

Generative artificial intelligence (GenAI) uses significant computing power resources to efficiently train models used to process data, which in turn generates content. Currently ChatGPT only uses data until 2021. However, regular training of the model to incorporate up-to-date data will significantly increase computational power. ChatGPT currently operates as a standalone service, but by incorporating it into a search engine, for example, it could see users jump from millions to billions in a short period of time.

Henry Azhder, an expert in artificial intelligence, said. The dramatic progress we’ve seen in generative artificial intelligence in the last few years is partly due to researchers using massive amounts of computation and data to train new models. Advances in generative AI have been met with much excitement and concern, but the environmental impacts of powering GenAI are often left out of the picture.

While researchers strive for more efficient and less computationally intensive models, the inescapable reality is that computing, local or cloud services, uses a large amount of energy generated by fossil fuels. Artificial intelligence will inevitably become an essential part of the global infrastructure, meaning that the resulting carbon emissions will only increase if we don’t accelerate the transition to renewables to fuel the manufacturing revolution.

What is the real cost?

Data center infrastructure is expected to grow by around 10% in the next few years. While the AI ​​boom is primarily responsible for significant growth as data centers struggle to meet demand. The increase in computing is likely to outpace the growth rate of the hyperscale community.

“Most companies today use GPUs (graphics processing units), which have incredible computing power, but the only way they work efficiently is to make the clusters as small as possible,” said Craig Hoffman, founder and CEO of Metro Edge Development Partners. be close to each other Data centers are not able to withstand the thermal load. That said, computing growth is estimated to be faster than the typical superscale community that has grown over the past three years. I envision exponential growth in the AI ​​space over the next few years.”

Lim May-Ann, director of the Fair Technology Institute, Access Partnership believes that awareness of the true cost of using AI needs to be effectively shared.

The challenge is in getting the general public to understand that accessing and using productive AI, especially services that are “free to use”, still comes at a cost, namely energy provision and absorption and therefore carbon emissions. Nasdaq has preliminary research showing that GenAI training alone consumes 1,287 megawatt-hours, or what they describe as the annual energy consumption of 121 American households.

Telecom companies will face pressure to upgrade their networks to prepare for increased demand. This will be challenging for countries that have just started or are in the middle of investing in 4G infrastructure and are not yet fully ready to invest in 5G networks.

Rafael Possamai, director of data center Bluebird Network, believes that while the investment is probably great, some of the traditional challenges of hardware procurement can be minimized.

The importance of investing in hardware to support the rise of productive AI depends largely on the scale and scope of deployment, the specific models used, and the overall strategy of companies. While the investment in hardware can be significant, it’s important to keep in mind that both Google and Microsoft have a strong foothold in cloud computing and infrastructure. This gives them the flexibility to allocate resources as needed and minimizes some of the challenges. However, as AI models continue to grow in complexity and size, investment in hardware to support their deployment is likely to be considered.

How does it affect the supply chain?

In fact, supply chain issues are also likely to be a cause for concern. said Craig Hoffman, founder and CEO of Metro Edge Development Partners. One of the largest GPU providers has stated that they are unable to access GPUs for lab testing today due to high demand. Organizations that can create a similar product with NVIDIA’s GPU (the leader in this space today) are likely to have similar supply chain issues with timely product delivery due to the rapid growth of AI. In order to meet the current demand, the supply chain must increase by more than 100% in deliveries to keep up with the rate of orders.

While companies have been evolving and preparing for the GenAI boom for years, they are in uncharted waters as artificial intelligence accelerates from being used to predict and diagnose network anomalies to improving customer service. The pressure to invest in infrastructure and storage, increase grid performance as well as balance environmental impacts makes it one of the most challenging times the industry has faced.

Ciena CTO Steve Alexander says that service providers must first address some essential architectural efficiencies and transform their networks into a simpler, more open and scalable infrastructure that will serve as a foundation for innovation in the era of distributed services and content.

“Essentially, you can’t automate what you can’t see, so automation and AI monetization won’t happen until the network infrastructure is fully visible, providing a source of truth.”

Environmental effects

The complexity of GenAI means that it consumes significantly more energy than traditional variations of computing. Currently, only energy and water (via data center cooling systems) are estimated to run AI models. But as they get bigger and more powerful, the impact on the environment is likely to increase.

David Hurst, group director for independent Australian data center provider Macquarie Data Centres. Artificial intelligence will have a profound impact on people, in every company and in every industry. This big explosion of artificial intelligence will lead to an explosion of data that we have never seen before. At the heart of it are the data centers that will power the AI ​​revolution.

Artificial intelligence presents a new sustainable challenge and opportunity for the industry. On the one hand, data-intensive workloads generated by AI training and inference further increase energy consumption. On the other hand, this technology itself has the ability to change the efficiency of the industry.

Operational efficiency and environmental sustainability go hand in hand. Efficient operation not only benefits the environment but also reduces costs. AI can make infrastructure more efficient through real-time insights into energy consumption and cooling, as well as lower energy efficiency numbers. Meanwhile, real-time insight into predictive maintenance can extend the life of data centers and reduce landfill.

The time spent on hundreds of millions, potentially billions, of simple mathematical calculations performed by artificial intelligence is significantly reduced by specialized hardware including GPUs and tensor processing units, while cooling methods such as liquid And directly to the chip is increasing. “Keep the high-density infrastructure that AI needs cool.”

The new law will come into effect in January 2024 in the form of the new EU Corporate Sustainability Reporting Directive, which will make businesses accountable for their environmental impact.

Legal image

Mandatory ESG disclosure and external verification will have a major impact on how companies report in the future. However, data centers are currently believed to be responsible for 1% of total greenhouse gas emissions.

Hirst believes the industry must now balance a massive growth cycle while improving its sustainability credentials.

Our digital footprint continues to grow and so does its impact on the environment. However, data centers are the most efficient and sustainable environments for implementing the digital economy.”

The responsibility of protecting the environment in the age of artificial intelligence is not only on the industry and its partners. Governments around the world can support the technology industry in achieving environmental, social and governance (ESG) goals by encouraging the adoption of sustainable practices through policy changes and financial incentives, including grants.

Thinking bigger, governments should prioritize investment in renewable energy. It provides the digital world with informed consumption and sustainable computing.

Organizations that use data center resources also play an important role. Superscalers, clouds and governments should look for data center partners that are flexible enough to adapt to new technologies such as artificial intelligence and maximize efficiency for sustainable implementation.

“As stewards of this technology, it is important to work with customers and governments to continue to foster this innovation in the most ethical and sustainable way.

#Artificial #intelligence #revolution #real #cost #telecommunications
Image Source :

Leave a Comment