The year 2030 bears a lot of weight – not only does it mark the end of the current decade, but it marks a major turning point for sustainability across several industries and nearly all countries. Of most relevance for this discussion are the carbon, emission and water neutrality goals that global companies have set themselves.
By 2030, Google intends to reach net zero, Microsoft wants to become carbon negative and Apple wants to make its entire footprint carbon-neutral, but these are all goals that were set as far back as 2019 and 2020.
A lot has changed since then, and nobody could ever have imagined just how much generative AI was going to blow up when ChatGPT launched in late 2022, let alone the impact that agentic AI continues to have.
Latest Videos FromTechRadar
With 2030 previously envisioned as some sort of a finish line, that deadline is now under immense pressure as hyperscalers expand at an incomprehensible pace just to keep up with cloud and compute demands. So can climate commitments that were designed before the AI boom survive the infrastructure demands it’s created?
AI has fundamentally changes sustainability progress
AI’s impact on the chips market is already well-reported. Omdia’s PC and tablet research director Ishan Dutt explained to me that an ongoing “capacity reallocation” is currently pushing memory makers’ “DRAM/NAND wafer capacity toward HBM and high-capacity DDR5 for AI data centers,” which is of course constricting supply for consumer devices.
Even without being familiar with the intricacies of this shift, consumers are already acutely aware of the ongoings due to sharp rises in PC, smartphone and storage prices.
What I’m more worried about is the long-term sustainability of AI-related expansions – while mounting local opposition is already calling hyperscalers and developers out over intense electricity and water consumption, this doesn’t begin to cover the longer-term impacts. Upcoming 2030 deadlines are a great opportunity to benchmark current progress.
I’ll explore four key areas of this in further detail with the help of Google’s Director of Sustainability for EMEA, Adam Elman: grid capacity, water consumption, the efficiency paradox and local socioeconomic impacts.
The grid is the first bottleneck
In our exclusive interview, Elman acknowledged that recent “hyper-growth” has strained progress toward 2030 goals, but for Google in particular, the targets were intentionally “ambitious” to “push the frontiers of what is possible in energy systems and data center operations.”
I questioned recent shifts in the trajectories of many companies’ sustainability reports, including but not limited to Google’s, such as spikes in energy consumption and emissions. Elman warned me that “the path to achieve these ambitions is not linear,” implying that companies will constantly make adjustments as the landscape evolves.
Google saw a 37% rise in annual electricity demand, but still succeeded on 100% matching with renewable energy purchases to maintain momentum toward its goal. Amazon also saw a 16% rise in its last-year total emissions, and a 34% rise in purchased electricity emissions.
These figures are widely reported in publicly available sustainability reports, but I asked Elman whether companies should be more transparent about the finer details. He agreed that being open “about the complexity of the energy transition” amid this turbulent AI era is “the right thing to do,” noting that this honesty can also help “drive the policy, market and technology developments needed to unlock an abundant clean energy future.”
The company also stresses the importance of distinguishing between global annual offsets and true local grid-level hourly matching so that customers, investors and policymakers can understand the challenges.
Global warming is highlighting water pressures
I also asked whether water self-sufficiency is a realistic goal amid rising global temperatures, to which Google’s response was overwhelmingly positive. Data center projects can follow one of many routes to reduce their impacts, including adopting closed-loop systems or using air cooling where that may be more suitable.
To relieve pressure on local water tables, Elman also told me that companies can opt to use recycled wastewater in closed-loop systems to get all the benefits of effective water cooling without consuming drinkable, fresh water.
Similar to energy reporting, Google supports local, site-level water reporting so that local citizens and utility companies can better understand the demands of individual campuses.
Efficiency drives total reductions
The reality is that emissions will inevitably rise as hyperscalers built out more and more data centers, but Elman told me that prioritizing absolute reductions or improving efficiency would be a false dichotomy: “True leadership requires prioritizing both absolute reductions and massive efficiency improvements.”
Efficiency is an important part of the equation, and decoupling compute growth from resource consumption means companies can understand progress.
I asked about installing smaller, hyperlocalized data centers as a measure to tame local opposition, which I’ll also explore below, but from an energy standpoint, Elman reminded me that larger facilities are generally more efficient because they can use large-scale cooling systems and support utility-scale energy agreements.
With ideas now starting to circulate around positioning data centers in the sea or as far as space just to improve thermal and power efficiency, Elman admitted that technical feasibility and practical implementation are two, very different things.
Google is now one of a growing list of companies looking at low-Earth orbit data centers, with Project Suncatcher observing an 8x energy per unit area increase compared with terrestrial solar panels. Prototype TPUs are set to arrive as soon as 2027, but as for wide-scale deployment, we’re likely years or decades off the real deal – if at all.
Responding to local opposition
Public opposition is also on the rise amid wide-scale and local sustainability concerns, with many US states and cities already having imposed temporary memorandums to ban new projects as they get to grips with policies.
Google’s EMEA Director of Sustainability told me that data center operators need to change how they’ve perceived, shifting from resource consumers into “supportive community anchors.”
Some measures that Google suggested including paying for grid upgrades to protect local households from rising energy bills and supporting water projects to offset consumption. It’s also not uncommon for companies to offer other local rewards to encourage citizens to accept their plans, like job opportunities, training schemes and other educational funding.
Is 2030 still a finish line, or the biggest test?
Ultimately, 2030 should not become an excuse for hyperscalers to withdraw or reframe the targets they set before the AI boom – instead, it should serve as an accelerant to encourage progress despite the challenges.
The technology has not made those earlier commitments obsolete, but AI has exposed how some earlier measures might not be so effective anymore.
Key to meeting these targets is openness and transparency around reporting for the purpose of policymaking, but it’s also clear that one route won’t solve the problem. Instead, success will come from endless components that will all add up to a meaningful result, be it driving efficiency, encouraging local support, responding to grid challenges, changing water habits or something else.
Whether or not the 2030 goal remains a finishing line is yet to be seen, but there’s clear optimism that it’s still a realistic goal for many.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
https://cdn.mos.cms.futurecdn.net/GcQXTy4NBXKeoop4V5WQnQ-970-80.jpg
Source link
