The $1T AI wake-up call
How ESG could have been used to forecast market disruption
Investors missed the warning signs. ESG factors, often dismissed and misunderstood, signaled the risk of AI disruption.
This isn’t just about one market drop. It’s a wake-up call. Ignoring ESG considerations creates risks and blinds investors to the next wave of disruption.
Why it matters: A single AI efficiency breakthrough crushed technology stock valuations. There were signs in ESG trends including energy efficiency, geopolitical strategy, and governance risks.
The AI boom fueled an already strong market frenzy:
Market growth has come from large-cap technology stocks.
The launch of Generative AI accelerated this recently, directly impacting technology companies and semiconductor manufacturers.
Nvidia stock is up over 2000% in the past 5 years, and TSMC and Broadcom are up over 170% in the same period.
Technology, financial firms, and energy companies are cashing in.
Two weeks ago, Oracle, OpenAI, and SoftBank announced a new $500B investment in US infrastructure and datacenters to support the AI boom.
But one breakthrough led to one trillion gone.
Chinese startup Deepseek announced massive efficiencies in GPU usage around the training of a transformer model at about 10% of the cost that Meta spent developing Llama and outperforming OpenAI’s ChatGPT.
This level of AI efficiency could mean fewer chips and less energy.
It also shows the overpriced nature of Western AI company valuations.
These efficiencies translated into a steep market drop, wiping out $1T of tech valuation in one session.
Yes, more efficiencies might equate to less energy and fewer chips needed, but it could open the door to steady use through increased demand. In economics, Jevons paradox is where efficiencies drive down costs and increase consumption. This is an argument for why long-term potential may not erode.
But, let’s focus. Could ESG have predicted this?
🍃⚡Environment
Efficiency gains were inevitable. Research predicted AI’s energy use would mirror what happened during the datacenter boom of the 2010s.
No hyperscale cloud provider pulled back sustainability goals after new AI advancements.
During the 2010s, internet traffic grew 16x, datacenters grew 9x, but related energy use only grew 10%. Efficiencies are the reason, and a paper published last year predicted this might happen again with AI.
Improvements in energy efficiency could offset some of the projected increase in power demand, as they did when data centres expanded in the 2010s4. More-efficient AI algorithms, smaller models and innovations in hardware and cooling systems should help5,6.
Regarding AI, efficiencies could impact everything from chip demand to reducing new construction of data centers to improved water and energy usage. Of course, chip development and energy usage will continue, but how much has yet to be seen because we don’t yet know the real efficiencies.
Please re-read that last sentence because if you think Jevons paradox will return the markets to whatever hockey stick growth for chips and energy investors were feeling, the reality is likely more nuanced as we’re in the early stages of AI adoption.
Consider that no hyperscale cloud provider has pulled back on their sustainability goals after Generative AI, so what would make anyone think that they weren’t banking on emerging efficiencies? Regardless of Deepseek, this was likely to hit at some point. While efficient training was developed in this case, efficient responses might be next.
⚫⚪Social
2017: AlphaGo beats China’s World Champion Ke Jiu.
China sets a national AI strategy to lead the world in AI by 2025.
2025: (checks watch)
In 2017, an AI model called AlphaGo beat world champion Ke Jiu. Like Sputnik sparked US interest and collective pride in the space race, this defeat led China to build a strategy to become an AI superpower.
Ke Jiu’s defeat was in May of 2017. By October of that year, the MIT Technology Review wrote about China’s plans for global AI dominance.
The plan calls for homegrown AI to match that developed in the West within three years, for China’s researchers to be making “major breakthroughs” by 2025, and for Chinese AI to be the envy of the world by 2030.
It is 2025, and that breakthrough is here.
Chinese developers mobilized around purpose at this scale. Why would investors be surprised? The social cohesion of China around this emotional loss and subsequent AI win should not have been overlooked, and it certainly shouldn’t be underestimated.
📈📉Governance
China’s centralized mobilization created an AI breakthrough.
Investor myopia missed the signs.
The mobilization of Chinese researchers under the thumb of a Communist government doesn’t point to quality governance. Still, when goals are set in such an environment, it appears that leaders can mobilize people and progress. Capitalism has entirely different sets of motivational methods, and with rampant polarization across the West, getting citizens pedaling in the same direction is a challenge.
Governance comes into play in another area, and it is the same place as every missed ESG consideration, with the people overlooking the issues. In this case, investors.
ESG is interconnected data
Having written a book about ESG, I keep returning to the acronym's interconnected nature. As I wrote about last week, I suspect we are entering a time when use cases will be written about company programs to showcase the value to stakeholders.
This is one use case where ESG considerations could have protected shareholder value.
Still, it is in the combination of these considerations that real insights surface and informed decisions can be made. Looking at complex issues in a silo or through just one additional lens leads to something interesting, but not actionable.
In the example of Deepseek, understanding the potential for AI efficiencies might have allowed investors to hedge their bets against chip or energy investments. While the Jevons paradox might explain a recovery and future growth, the realities of that recovery are hard to see while you are on the possible upswing towards increased consumption. Understanding China's likely and now-arrived announcement around their 2025 goal would have added to the evidence that something was coming.
If you are looking for other risks that could disrupt the AI boom, I suggest you consider how the DEI pullback could negatively impact any stakeholder-centric AI model. As companies make AI advancements, the need for domain expertise in this area will lead to more risks unless responsible AI principles are followed. Remember, AI adjustments have been made before.
We have a shorthand framework in ESG to understand all of these nuances if we admit it is useful. But, as of now, most investors aren’t adjusting their strategies and don’t see this announcement impacting tech stocks over the long term.
The long-term view is an ESG one, yet investors still ignore its analytical power. AI will keep disrupting, but those who connect the dots will see the next shock coming.
AI efficiency wasn’t a surprise. ESG data showed the trend.
A 2017 roadmap made geopolitical AI dominance clear.
The next disruption is already forming in ESG signals.
Don’t just react. Use ESG insights and adapt.