Artificial intelligence is known for its seemingly insatiable appetite for energy. But some tech leaders and analysts are questioning the extent of AI’s footprint going forward, saying innovations in the sector could help offset rising energy demand.

There is certainly no shortage of forecasts detailing the ominous rise in AI’s energy consumption. One commonly-cited report by Climate Action against Disinformation, an association of environmental groups, suggests that AI could drive up global emissions by 80%.

Another estimate by researcher Alex de Vries—which he described as a worst-case scenario—suggests that Google’s AI alone could eventually rack up as much annual electricity demand as the country of Ireland.

Such predictions present a substantial challenge for the relationship between computing and the climate in the coming years.

However, others see reason for optimism. In a Salesforce survey of about 500 corporate sustainability professionals, published Wednesday, nearly half were concerned about AI’s potential negative impacts on sustainability efforts. Meanwhile, almost 60% thought the benefits of AI would offset its risks in addressing the climate crisis.

Microsoft founder  Bill Gates recently weighed in on the subject, urging governments not to go “overboard” on concerns about AI’s energy footprint, and suggesting that the technology  could actually drive a reduction in global energy demands.

Putting the AI boom in context

The rise of AI should be considered in a broader perspective, according to Charles Boakye, U.S. sustainability analyst at Jefferies. He noted that relative to other industries, the technology still makes up only a fraction of global power demands.

“Data centers—the engines powering everything from AI to traditional computing to cryptocurrency—currently account for about 2% of global electricity consumption. Of this, AI accounts for roughly 0.5%,” Boakye said.

In terms of emissions, statistics last year from the International Energy Agency showed in total, data centers and transmission networks are responsible for 1% of energy-related greenhouse gases. For comparison, the oil-and-gas industry contributed just under 15% of emissions in 2022.

Looking ahead, the intergovernmental organisation said that by 2026, demand for AI is expected to increase ten fold compared with 2023. Even so, IEA data showed that it will still only account for roughly an eighth of total data-center electricity consumption.

“So in terms of AI demand, we’re talking about a small piece of a small piece,” Boakye said, noting that other areas of electrification, such as electric mobility, traditional data center growth and the industrial transition, will ask much greater questions of the power sector.

Charting efficiency trends 

Demand is difficult to predict. But recent trends show that in practice, a rise in demand for computing power rarely correlates one for one with a rise in energy consumption, according to climate researcher Jonathan Koomey.

“Between 2010 and 2018, global data centres saw a 550% increase in compute instances and a 2,500% increase in storage capacity. This compares with just a 6% rise in electricity use,” he said.

Koomey, previously a visiting professor at Stanford, Yale and Berkeley, is best known for his work studying long-term trends in the energy-efficiency—now known as Koomey’s Law—highlighting the propensity of computing technologies to become more efficient over time.

AI may be a different animal, with some estimates suggesting large models such as ChatGPT use 10 times more energy than a Google search. But this is unsurprising for a relatively new technology, Koomey noted. In most areas of computing, energy demand tends to spike before levelling off as efficiencies gather pace, he said, noting that forecasts based on this inflection point will often be misguided.

Koomey cited a brief moment in the mid 1990s when internet data flows doubled every hundred days—a statistic that led to overinvestment in networks in the following years and 97% of fibre capacity sitting unused.

Similar efficiency trends can be seen in the development of AI, Jefferies analyst Boakye said. Google’s new TPU processors, for example, are more than 67% more efficient than in 2022, and the energy used to train OpenAI’s ChatGPT models has gone down by 350 times since 2016, he noted.

“It’s in their best interest, and in the interest of their business models, to increase that efficiency,” the analyst said.

Going for gold in the AI Olympics 

If any company will have a say in the future of AI, it’s Nvidia . The U.S. technology company designs roughly 80% of the world’s specialized AI chips. Its next-generation GPUs, known as Blackwell, are touted to be 25 times more energy efficient than current iteration Hopper, while offering 30 times more computing power. Blackwell chips are slated for release later this year.

So far, the progress appears consistent, with Hopper 25 to 30 times more efficient than Nvidia’s previous generation of chips. Overall, the company said it has experienced a 45,000 times improvement in GPU energy efficiency over the last eight years.

Nvidia added that software optimisation also plays a big role in increasing the energy efficiency of its products.

“Once a platform has been launched, we will make it more efficient in a single year,” the company said. In the year following its launch in 2022, Hopper became two times more efficient after taking part in MLPerf, otherwise known as the Olympics of AI, in which tech companies compete and collaborate to drive improvements in the speed and efficiency of their models.

Part of this optimization involves taking large, energy-intensive AI models—such as ChatGPT—and refining them to perform more specific tasks. On July 18, for example, OpenAI launched a  smaller, smarter and more energy efficient  version of its previous GPT model, known as GPT-4o mini. Koomey sees these more “lightweight” models  driving demand in the future .

“I’m not convinced large-scale AI has a good business model at this point, despite them driving all the investment. The slam dunk machine learning usually involves smaller, more efficient models, focused on a specific task. For me, this is where most of the business value lies,” he said.

Training and education will be essential for getting a more targeted and efficient experience with AI, helping to narrow the gap between businesses and their sustainability goals, said Suzanne DiBianca, Salesforce’s chief impact officer.

According to Nvidia, another way of alleviating AI’s impact is directing workloads—particularly the more energy-intensive jobs of training AI models—to regions with more abundant resources, ideally renewables which are set to keep a lid on electricity emissions in the coming years.

One company making strides in this space is NexGen Cloud, which builds renewable-powered data centers in areas with untapped energy resources.

According to co-founder and CSO Youlian Tzanev, a large portion of AI workloads don’t need to be performed close to traditional logistical hubs like London or New York, and can be powered from more remote areas in countries like Canada and Norway with excess hydropower.

“We have significantly more power than people believe. The power just isn’t reaching the grid in many cases and is going to waste, and so that is where we focus our efforts,” Tzanev said.

In the U.S., Crusoe Energy Systems offers another example of startups finding innovative ways to power the AI boom. Crusoe’s modular data centers are designed to run on excess natural gas produced at oil wells, achieving a 99.9% methane reduction in the process.

Common forecasting pitfalls 

Projecting an outlook for AI’s energy demands is far from straightforward. In its midyear electricity update, the IEA noted that estimates exhibit a wide range of uncertainty, with some analyses following overly “simplistic” extrapolations.

For instance, the organization said certain studies make the mistake of assuming data center operators build all the facilities for which they apply to utilities. Given that several applications can be made for each new data center, this can lead to a multiplication of estimates.

Within data centers themselves, it is tempting for forecasters to imagine computers working flat out around the clock, Koomey said. In practice, GPUs usually operate on much less than their full power capacity, he added.

Based on modeling carried out by Nvidia, GPUs on average tend to run on less than 70% of their potential power. One particular function, known as Multi-Instance GPU, enables workloads—and therefore energy consumption—to be split into seven distinct components, with each able to function independently.

In addition, the company noted the role of substitution effects, in which traditional computing workloads are transferred onto AI platforms and subsequently performed more efficiently—an aspect that can be easily overlooked.

Forecasts can also  conflate local data-centre developments with broader energy demands , Koomey added, noting that the most common estimates for electricity demand come from local utility companies.

In the U.S. as a whole, electricity use actually fell in 2023 compared with the previous year, according to the U.S. Energy Information Administration. In March 2024—the most recent month of available data—total demand reached 306 billion kilowatt-hours, down from 317 billion kWh in the year-prior period.

“I worry that people are jumping on the explosive demand-growth train before really understanding what’s going on,” Koomey said. “If you cluster data centers in certain places you’re going to see some local power constraints, but that doesn’t necessarily mean that AI will be a key driver of electricity use more broadly.”

 

Corrections & Amplifications undefined Nvidia said it has experienced a 45,000 times improvement in GPU energy efficiency over the last eight years. An earlier version of this article incorrectly said the company developed its first GPU eight years ago. (Corrected on July 24)

Inflation unexpectedly heated up in the eurozone this month, presenting a fresh challenge to policymakers looking for signs that eurozone price rises are easing sustainably.

Consumer prices were 2.6% higher on year in July, picking up pace from in June, according to EU figures released Wednesday. That defied economists’ expectations for a slight decrease in inflation over the month, and leaves the rate further from the European Central Bank’s elusive 2% target. Inflation heated up over the month in Germany, France and Italy, the eurozone’s three largest economies.

Core inflation, which strips out the often volatile effects of food and energy prices, meanwhile stayed stable, against expectations for a slight decrease. But services, a key focus for policymakers at the European Central Bank, did fall slightly.

That decline makes a cut to interest rates the most likely outcome in September, said Franziska Palmas, an economist at Capital Economics.

Still, the higher headline rate may give pause to the central bank as it mulls its next steps on the direction of interest rates in the 20-member eurozone.

The ECB cut interest rates in the currency union for the first time in five years last month, but has refused to be drawn on when and how quickly it will continue to lower rates from their current heights. Markets still expect a second cut to rates at the bank’s next policy meeting in September, but with less certainty than earlier in the summer.

“Until services inflation falls more significantly, the ECB is likely to continue to ease policy only slowly,” Palmas said.

All eyes will rightly be on the Federal Reserve’s interest-rate decision Wednesday. But the meeting of another central bank across the Pacific will be quite consequential too.

While Jerome Powell is pondering whether to cut rates now for the first time in more than four years or perhaps wait a couple more months, Kazuo Ueda , his counterpart in Japan, is considering whether to do the opposite . After the Bank of Japan exited its negative-interest-rate regime in March, investors are looking for more tightening to come.

The Bank of Japan raised its short-term interest rate from minus 0.1% to a range of around 0% to 0.1% in March, the first increase in 17 years. Japan’s consumer prices rose 2.8% year-on-year in June, which was off from inflation’s peak pace last year, but still well higher than Japan is accustomed to.

In a possible preview of what awaits markets if the Bank of Japan leans hawkish, the Japanese yen has risen sharply in the past few weeks, appreciating 5.2% against the dollar this month from a multi-decade low. That has likely contributed to market turmoil in other markets around the world over the past couple of weeks, including the selloff in global technology stocks, as the yen, with its low interest rates, is a favourite funding currency for traders .

Hedge funds have ramped up their short bets on the yen in the past two years but could exit their positions pretty abruptly. Leveraged funds have slashed their net short position in options and futures against the yen by half in the two weeks ended July 23, according to data from Commodity Futures Trading Commission via CEIC. That is equal to a nominal value of $4.6 billion.

But the market is divided over whether a Japanese rate increase could come as soon as this week. There is a 41% probability that the Bank of Japan could raise rates by 0.15 percentage point, inferred from pricing of overnight indexed swaps, according to Bank of America.

At the meeting this Wednesday, the Bank of Japan is also expected to outline plans to unwind its portfolio of $3.8 trillion in Japanese government bonds, likely giving a further boost to long-term rates. Japan’s 10-year government bond yields have gone up 0.44 percentage point to around 1.06% this year as investors expected higher rates.

The country was swimming against the tide over the past few years by staying put on its ultra-easy monetary policies when most major central banks were raising rates. The Bank of Japan will likely be more cautious going in the opposite direction: A sharply higher yen could be punishing to exporters , and policymakers in Japan live in perpetual fear of returning to deflation. Any surprises in the BOJ’s pace as it normalises policy could still rattle financial markets.

Over the longer term, a narrowing interest-rate differential between the U.S. and Japan could shift the pattern of investment flows. Japan had the equivalent of $4 trillion in foreign portfolio investments at the end of 2023, according to official data. That includes both companies and individuals which are scouring the globe for higher returns. Some of them might bring their money back to Japan if assets at home are generating higher yields, especially if the yen is getting stronger.

U.S. 10-year government bonds still yield around 3.1 percentage points more than Japanese ones, but that is already down from a 4.2-percentage-point gap in October. If that gap keeps narrowing, it could mean tighter financing conditions in markets around the world, which have long looked to Japan as a steady buyer.

It is rare for two of the world’s largest central banks to be at major turning points in their long-term policy settings at the same time, and rarer still for them to be moving in opposite directions. The consequences could be far-ranging and unpredictable. Investors around the world will have to get used to paying more attention to what is happening in Tokyo, not just Washington, D.C.

Whenever people talk about the dangers AI holds for the workforce, they usually have one thing in mind: technology stealing jobs. But artificial intelligence poses a much more subtle threat than that—one that will have consequences for business unless we address it.

Simply put, the way we’re handling AI is keeping young workers from learning skills.

For more than 12 years, I have been studying how work changes as a result of intelligent technologies like robots and AI. Across a number of industries, I’ve seen the same thing over and over: This new, sophisticated technology makes it easier for experts to do their jobs. Seasoned surgeons can operate more quickly and efficiently, for instance, when they use robots in the operating room.

But the efficiency comes at a cost. The technology allows experts to do more, independently, so they don’t need younger, less-experienced workers to help them out anymore—so those novices are left without mentors to teach them the skills they need to do their job. Looking at operating rooms again, it takes two people to perform most complex procedures with traditional tools. The senior surgeon generally provides “exposure” by retracting tissue while the resident does what most of us think of as surgery—incisions, suturing and so on. Residents are on task the entire time. Focused. Learning.

Now the residents mostly sit around during operations and watch veteran surgeons get the job done thanks to help from a robot. Limited work. Limited learning.

As learning opportunities like these are lost throughout more industries, the results could be profound for both individual workers and the economy. We are sacrificing skill building and human bonds of mentoring on the altar of productivity. No matter our role, tenure, occupation or industry, if we can’t collaborate with someone who knows more, we’re not going to learn effectively, and we won’t be able to keep up. And our organisations will struggle where they might otherwise race ahead—because workers won’t have the deep knowledge they need to innovate and step into senior roles.

Turning history on its head

We have decades of research showing that this situation is the opposite of what we want. We build skill by collaborating across the expert/novice divide, so novices get to see the work, help out at the edges and earn the privilege of doing more next time.

Now that mechanism is being lost. My observations, combined with primary data from other field researchers, show a destructive dynamic at work, across a range of industries. In industrial-process engineering, I have seen experts use software to do modeling on their own, instead of involving a junior engineer. In warehousing, I’ve watched area managers rely on dashboard analytics to understand staffing and process flows, instead of uncovering those things collaboratively with less-experienced line leads and workers.

My collaborator Callen Anthony at New York University found that junior investment-banking analysts were being separated from senior partners as those partners started to use algorithms to help create company valuations for mergers and acquisitions. Junior analysts—instead of collaborating with the senior partners as they had before—essentially just pulled data for the algorithms to use in their valuations.

The rationale for this arrangement was twofold: reduce errors by junior people in sophisticated work and maximise senior partners’ efficiency. Explaining the work to junior staffers pulled partners away from higher-level analysis.

This setup produced short-run productivity improvement, but it moved junior analysts away from challenging, complex work, making it harder for them to learn the entire valuation process and diminishing the firm’s future capability. Junior bankers become senior bankers, after all.

Less time at the table

One of the most striking examples of the widening skill gap is surgery. I observed hundreds of procedures at some of the top teaching hospitals in the country, where robots deeply reshaped how work was done. Surgery, as I said, used to take four hands; minimally invasive surgical robots can supply three, all controllable from a single console. They make things so much easier for surgeons that the million-dollar tools have become the de facto standard for many complex procedures.

Most important, robots make it possible for surgeons to perform operations solo, no residents needed. And, since residents are slower and make more mistakes than an experienced surgeon would, those surgeons are opting to cut residents out of the action. Before, residents might operate for four hours during a 4½-hour procedure. In my nationwide data, their robotic average time hovered in the 10- to 15-minute range. And residents got less operating time in 88% to 92% of cases.

In this situation, we end up with much-less-capable surgeons. My data shows that many newly minted surgeons struggle mightily when they get their first jobs—not just because they don’t have robotic skill, but because their failed quest to learn robotics took so much effort they lost key learning opportunities in other procedures and practice areas, from ureteroscopy to kidney stones to vasectomies, that they would be expected to handle in most new surgical jobs.

The long-term loss

The consequences of poor training go beyond day-to-day competence. Consider what happens to the culture of a hospital when it loses healthy expert/novice collaborations. Less teaching and learning, to be sure, but also more-limited career advancement as experts advocate less for trainees. What about hospitals’ ability to innovate in surgical practices? Limits there, too, as discoveries made by colleagues get tamped down by increasingly focused, efficient, expert-driven surgical performance. The ability to service skyrocketing surgical demand? In the short run, you serve more patients, but in the medium term you scramble to keep up as the pool of new talent dwindles.

Of course, different organisations, industries and professions in different places will feel the pinch on different time scales. They will also compensate in different ways. But in general, organisations will not sense the problem directly: Instead, they will incrementally accumulate a larger cost base—in areas such as (re)training and reduced billable or applied time—and build a bureaucracy to manage this skills gap. At law firms, new attorneys might take longer to ramp up to normal caseloads, while senior attorneys would have to spend more non billable time to handhold them.

Now imagine the consequences of similar skills gap across all types of companies, throughout the economy. Without a firm, immediate correction, this is what we can expect. This is our trillion-dollar skills problem.

A way forward?

Solving the problem is vital, but how should we do it? My collaborator and I found evidence of one approach that can work.

Remember, the problem right now is that senior workers are learning new technologies, such as robotic surgery, that make junior workers unnecessary. In our research, though, we found cases where junior and senior workers teamed up to learn about new technologies together .

By working closely with seniors in this way, the juniors didn’t just learn about the new technologies, they ended up collaborating with seniors on other aspects of the job. Since the older and younger workers were figuring out how the tech worked, they also needed to figure out how to integrate it into vital day-to-day tasks. So, the novices got to see firsthand how those jobs were done while performing actual work.

For instance, in my research, I saw some residents and senior urologists team up to learn robotic techniques in live surgical procedures. In those cases, the residents got much more actual hands-on operating time than residents who mostly just watched robotic procedures—10 times more. And the quality of that time was far better: Expert and novice were jointly figuring out how to use the tech, just as they had a patient on the table.

Granted, this process isn’t easy. In our research, we found that these collaborations often failed. But when they did work, they were powerfully effective. We need more companies to take the chance and implement this strategy, to figure out how to make it most effective and serve as examples.

It will not only help close the skills gap, it will give old and new workers a new sense of purpose on the job—through strengthened relationships. Research shows very clearly that we get motivation for our work when it builds trust and respect with those who share our values. Progressing to more competence therefore involves questions of the heart, like, “Have I earned this expert’s trust and respect?” or “Does this novice look up to me?”

We often treat these issues as unconnected with hard-nosed skill and results, when they are a core part of why we try at all in the first place. They are the animating force for the journey.

Matthew Beane is an assistant professor at the University of California, Santa Barbara, and author of  “ The Skill Code: How to Save Human Ability in an Age of Intelligent Machines.”

Sick of managing people? Maybe you should stop.

So many of us stumble into being the boss, or raise our hands because it feels like the only way to get ahead. We’re attracted to the cachet of the title, the promise of more money or the comfort of having a ladder to ascend.

Then come the performance reviews to write, the team drama to adjudicate, the meetings to attend . The job keeps getting harder. Managers oversee nearly three times as many people today as they did in 2017, according to data from research and advisory firm Gartner . Nearly one in five managers says that, given a choice, they’d prefer not to oversee people.

“That’s what we call buyer’s remorse,” says Swagatam Basu , a senior director in Gartner’s human-resources practice.

You can switch back. And your company might be amenable. More are “unbossing” their workplaces by shrinking middle-management layers .

The trick is figuring out a way to maintain your pay and influence. In some companies, the number of people you manage is a proxy for your power. Others now use special individual-contributor tracks, meant to ensure that technical experts have a set path to climb.

You might have to give something up. Making the shift could still feel like a relief.

“It was like, oh, I don’t have to deal with the people issues,” says Suzet McKinney , an executive at Sterling Bay, a Chicago real-estate company. She’d served in leadership positions before. When she started her current role in 2021—no pay cut required—she figured she’d eventually hire direct reports and build out a team. Then she realized she didn’t miss it.

“Managing people would be more of a distraction,” she says.

Making the ask

Dennis Henry , an engineering director overseeing about 45 staffers, was hungry to move to the next managerial rung at software company Okta last year. Then his supervisor explained that would mean even less time to do the technical work he loved. It made the 38-year-old wonder: Did he want to be a boss at all?

“What would hurt more?” Henry asked himself. Giving up managing or giving up coding? The latter felt unfathomable.

He pondered what he’d want if he left management entirely and became an individual contributor, ranking priorities. Maintaining his base salary—just shy of $300,000—was tops. He told his boss that he was happy to stay in his current role if a new opportunity didn’t pan out.

“You have to be ready to hear ‘no,’ ” the Orlando, Fla., resident says.

He got a yes: The company created a new job for him and preserved his pay. After 15 years as a manager, carving out a new kind of authority has been a transition.

As a boss, “I could just say, ‘Do this,’ ” he says. Now he spends more time amassing evidence for his ideas, making his case.

“It is so much harder to convince people that something is the best option,” he says.

The stress of managing

Jenny Blake ’s mental health took a dive after she was promoted to team lead at Google at age 24. She felt stressed and emotionally drained, deeply responsible for her team but beholden to decisions from above, like a department reorganisation ordered up by executives.

A 2024 survey from SHRM, a lobby for human-resources professionals, found that 40% of respondents said their mental health declined when they took on a managerial or leadership role.

Blake switched to an individual contributor job, spending several years rolling out new programs she felt had a much bigger impact than her management. Now an author and speaker focused on careers and business, she recommends broaching the transition conversation by laying out your unique strengths and how they can better serve the company in a new role. Don’t dwell on your distaste for managing people.

Want to ensure the shift isn’t a demotion? Make sure you’re staying close to parts of the business that are directly tied to revenue, she says. Build your reputation externally, speaking at conferences and publishing papers.

“Become an industry expert,” she says.

The reality of switching

Just because a company touts opportunities for individual contributors to grow doesn’t mean you’ll be able to rise to the top unimpeded. A former consultant at a professional-services firm told me that partners who didn’t have their own teams were treated like second-class citizens.

At Launch Potato, a digital-media company based in Delray Beach, Fla., the individual-contributor track tops out several levels below the executive level. Even on the lower rungs, managers have the opportunity to make higher salaries and bonuses than commensurate individual contributors, says Kristopher Osborne , the company’s senior vice president of talent.

“You are getting paid a premium to deal with a lot more issues and challenges,” he says of managers. “People have to be realistic.”

He recommends ambitious individual contributors show they’re bringing leadership to the company in different ways. Can you run strategy initiatives, coach teammates or get swaths of the organization on board with new initiatives?

Letting go

In a previous job, Sheri Byrne-Haber liked managing people and being a “one-stop shop” for her 20-person digital-accessibility department, even as the workload ballooned. So when her boss suggested splitting her role in two, she initially said no.

She reconsidered when performance-review season arrived. She had to write 19.

The company hired a new counterpart for her, charged with managing, and Byrne-Haber focused on strategy. Letting go was harder than she expected. It took her three months to unsubscribe from all the manager-only Slack channels, email lists and meetings she had been looped in on. When colleagues reached out with questions, she’d pause to determine whether the queries were still related to her responsibilities. If not, she forced herself to forward them to the new manager, even when she knew the answer.

“It felt awkward,” says Byrne-Haber, now at work on her own startup. “But that’s not my job anymore.”