AI boom: Will there be an energy crunch?
Theodora Lau
American Banker Top 20 Most Influential Women in Fintech | Book Author - Beyond Good (2021), Metaverse Economy (2023) | Founder - Unconventional Ventures | Podcast - One Vision | Advisor | Public Speaker | Top Voice |
With all the excitement around artificial intelligence, will demand ever outstrip supply?
That's been on my mind lately, especially given all the forecasts that I've been reading. And yes, winter is coming —?so will the strain on our power grid for those living in colder parts of the country.
Of the 8,000 data centers that exist globally, about a third are in the US, compared to 16% in Europe and almost 10% in China. And Northern Virginia (aka data center alley), with close proximity to the nation's capital, is the largest data center market in the US.
According to a Bain study , "By 2028, utilities would need to increase annual energy generation by between 7% and 26% above the 2023 total to meet projected demand." That's staggering, to say the least.
And that's just one piece of the puzzle; we still need to account for the water consumption as well. The figures are a bit hard to come by though. But it does give you a sense of the magnitude of the challenge.
No wonder we start seeing big technology companies signing deals in search of cleaner and more efficient alternatives.
Watch this space. It's simply fascinating.
What I am reading ...
[1] Using AI to enhance monitoring or ML/TF risks
The Hong Kong Monetary Authority (HKMA) has been a bit busy with AI lately. The regulatory is taking steps to support the use of AI in enhancing the monitoring of money laundering and terrorist financing (ML/TF) risks within authorized institutions.
They are also planning to unveil its inaugural policy statement on the use of AI within the financial industry across sectors such as investment banking, trading, and cryptocurrencies, during Hong Kong FinTech Week .
领英推荐
[2] 100 million AI agents (and counting)
Will NVIDIA someday be a 50,000 employee company with a 100 million AI assistants? This might be hard to imagine now but NVIDIA's CEO Jensen Huang envisions a future where the tech giant will mass deploy AI assistants across every division to improve output.
[3] AI video tools
Not to be outdone by competition, Adobe released the Firefly Video Model, which can generate video from text prompts, to compete with OpenAI's Sora (introduced earlier this year). TikTok owner ByteDance and Meta Platforms have also announced their video tools in recent months.
What did we just say about energy consumption again? :)
_____
In this week's One Vision podcast ...
As we often say, look East for fintech innovation. Luckily, you don’t have to get on a plane this time. In this week’s One Vision —?FinTech Fuse! Join Sanjeev Kumar —?CEO at WhiteSight , and myself, as we chat about the growing fintech ecosystem in Brazil, Asia, and the Middle East, challenges and opportunities in fintech globalization, and the importance of customer trust. Can we build global fintechs or will they stay local and specialized? How will the fintech ecosystem evolve for small and medium-sized businesses? Are fintechs delivering on customer trust?
Tune in and find out!
Self Employed Independent Financial Consultant
1 个月Theodora Lau As "Kamunism" and the war cycle reach new levels, neither ‘Drill Baby Drill’ nor the current slowdown in demand should deter oil bulls. https://themacrobutler.substack.com/p/why-neither-kamunism-nor-drill-baby
Lead Future Tech with Human Impact| CEO & Founder, Top 100 Women of the Future | Award winning Fintech and Future Tech Influencer| Educator| Keynote Speaker | Advisor| Responsible AI, VR, Metaverse Web3
1 个月Not only NVDIA, but all companies will have less employees and more - millions of agents soon.
Theodora Lau, aI's impact on finance is massive, huh? Looking forward to hearing more about those video tools. What’s your take on them?
Top Global Fintech & Tech Influencer ? Trusted by Finserv & Tech Global ? Content & Influencer Services ? Advisory for Digital Transformation ? Speaking ? [email protected]
1 个月We have double standards. Media are so excited about the unsatiable appetite for AI compute. Your thematic focus today reminded of an article i wrote back in feb (seems like a life time) Digital Dilemmas around Sustainability: Why Bitcoin not GenAI https://link.medium.com/V9NJVim1JNb
Leader, Gen AI & Cloud Evangelist, Storyteller
1 个月Great post. In our work with customers I actually was cheeky and have been using “reduce, reuse, recycle” as a theme. The reality is this is an energy intensive workload but there are ways to limit that… Reduce = smaller more targeted models. These require less infrastructure and can often deliver better results when trained on your data. Reuse = instead of proliferating models often you can use a model for many different departments or tasks. The trick here is to plug in different data sets based on different needs but having one model serve many. Recycle = probably the biggest one. When we started everyone thought they would need to build their own models. By leveraging open source you can often remove or greatly reduce the energy consumption of training. Either using off the shelf, off the shelf augmented with your data (RAG) or doing minor fine tuning. This is something we all need to work together to solve. #Iwork4dell