Can tech be sustainable?
The Profit Mandate
The U.S. tech industry is no stranger to environmental criticism. The recent boom in AI data centers is the latest in a long list of headlines highlighting the environmental consequences of Silicon Valley’s breakneck pace.
The industry’s well-documented extractive processes are often viewed as inevitable concomitants of its hyperscale growth. Is there a future in which tech is able to exist within a sustainable framework?
This article will explore this using the following definition for sustainability: the capacity of a system to maintain its function indefinitely without depleting the resources it depends on or damaging the environment beyond repair.
It’s important to first understand the U.S. political and economic context in which tech operates.
The U.S. economy and its constituent corporations operate under a neoliberal, capitalist framework, defined by the theory of shareholder primacy. First articulated in Dodge v. Ford Motor Co in 1919, the Michigan Supreme Court established a legal precedent that businesses are organized and carried primarily for the profit of their shareholders.
Economic stagnation in the 1970s led to the rise of neoliberalism and an emphasis on deregulation. Buckley v. Valeo in 1976 established election spending as constitutionally protected free speech, and Citizens United in 2010 removed limits on political spending.
Meanwhile, the Great Depression catalyzed a shift towards greater transparency into company health and an emphasis on future growth, measured through economic indicators focusing on a company’s projected future revenue.
The results are two-fold: corporations gain immense influence over U.S. politics through their financial donations, and they are legally bound to pursue profit above all else to ensure survival.
Under this system, pursuing profit above all else while shaping policy to favor your interests is both rational and necessary.
While the vast capital and hypervisibility of tech shape much of the digital, political and economic landscape, tech is bound by the same limitations and incentives as all U.S. industry.
Despite messaging about accelerating our progression into a futuristic utopia, Silicon Valley is about industry first and technology second.
The creation of the internet and software revolutionized the profitability model through extreme scalability, enabling exponential increases in revenue as the marginal cost of software development remained relatively flat.
A key example is the Software-as-a-Service (SaaS) model, which was pioneered by Salesforce. Salesforce created an initial subscription-based Internet application and then tailored it for various customers. This allowed for ridiculous profitability, enabling tech to write the rules of how they, and by extension, the rest of the nation's industry, should behave.
This is critical as current environmental policy — one of the few guardrails against rapid, hyperscale extraction — is subject to lobbying by corporate interests.
Despite ample manpower and resources, slower but necessary investments into sustainability aren’t prioritized because they are less profitable.
Extraction and Obsolescence
The tension between sustainability goals and profit incentives is well illustrated by the hardware lifecycle. Every technological device begins with the extraction and processing of raw materials.
Rare and valuable metals are often mined in mineral-rich countries by foreign entities, who profit on the land while local citizens remain impoverished.
Mines in the Democratic Republic of Congo have been linked to the supply chains of Apple, Samsung, Tesla, Microsoft and Dell according to GIJC. The country is rich in cobalt, gold, tungsten, tin and tantalum, all of which are non-renewable resources but essential components for circuit boards and batteries.
Mineral mining driven by the technology industry has fueled political conflicts and led to the use of child labor and other human rights abuses.
Following the extraction of raw materials, the refining processes of these minerals are water-intensive and generate high volumes of toxic waste and carbon emissions.
And when technology breaks or becomes obsolete, it’s discarded or recycled as electronic waste, or e-waste. E-waste contains substances that pose health and environmental risks when not recycled properly, which the majority is not.
To continually drive profit, the business models of hardware industry giants are predicated on planned obsolescence.
Planned obsolescence is the strategy of creating products with predetermined lifespans to force continual consumption.
Take Apple, whose 2026 Q1 earnings report states that hardware sales account for about 75% of annual revenue.
A definitive example of planned obsolescence was Batterygate, in which iOS updates were used to deliberately throttle the battery life of older models.
Apple has also used other tactics to force consumers to continue purchasing, such as installing proprietary pentalobe screws to prevent users from repairing/replacing parts of their own devices, according to iFixIt.
Apple’s sustainability reports focus on increasing the usage of recycled materials, reducing waste sent to landfills, and stewarding water resources.
These efforts are largely incompatible with the associated costs of producing more units while rendering older technology unusable.
While sustainability efforts focus on improving specific phases of the supply chain, net emissions and extracted materials continue to grow in tandem with units of hardware produced.
Thirst for Compute
The demand for endless growth also strains energy and natural resources, as seen in the expansion of data centers.
The aim of data centers is to provide compute, which is the processing power and hardware needed to execute software tasks.
Initially, data centers started in-house, with companies housing their own IT infrastructure in closets and basements.
The dot com boom accelerated the construction of dedicated server rooms, which led to larger infrastructure demands after the rise of cloud computing.
Hyperscalers like Amazon Web Services, Google Cloud and Microsoft Azure began building huge data centers to rent out processing power as needed.
With the advent of AI, the high demand for machine learning workloads has once again placed burdens on available compute, and tech hypergiants have been pouring money into building the necessary infrastructure.
Data centers require huge amounts of electricity to function, with the majority powered by fossil fuels, and drink millions of gallons of water to prevent overheating and damage to their servers.
However, there are more sustainable alternatives to water cooling, such as immersion cooling (where hardware is submerged in a reusable coolant), air cooling (using air conditioning vents) and free cooling (using outside cold air, but necessitates colder climates).
The advantage of water cooling is that it is cheapest and location-agnostic, allowing data centers to be built rapidly and pervasively.
Many data centers are also built in areas with moderate to high water stress. Most critically, the water consumed is nonrenewable, as it evaporates as steam and is lost to the local watershed.
This tension between sustainability and profit in the context of data centers is evident in Google’s sustainability reports.
In 2020, Google set some climate goals, with their most ambitious “moonshot” being to become carbon-free by 2030, primarily through designing more efficient data centers and advancing carbon-free energy. In Google’s 2025 sustainability report, while data centers have become more efficient and their energy emissions have decreased, absolute emissions have risen by 51% since 2019.
Due to AI driving demand for Google Cloud servers, Google has doubled its infrastructure spending on new data centers.
AI is becoming one of Google’s most profitable products.
Gemini has been integrated with Google Search and Google has entered into $460 billion in cloud contracts, which must be fulfilled through constructing new data centers at the risk of losing market share, according to Tikr.
As stated in the most recent report: “While we remain committed to our climate moonshots, it’s become clear that achieving them is now more complex and challenging across every level. Even though we’ve successfully reduced our data center energy emissions, supply chain emissions have risen. Additional external factors—largely outside our direct control—are converging to create significant uncertainty, including the slower-than-needed deployment of CFE technologies, AI’s energy demands, policy uncertainties, resource-challenged markets, and more.”
Again, slower investments in more environmentally friendly technology give way to meeting profit benchmarks more quickly. In the report, the emissions sacrifice is a foregone conclusion; the commitment only existed when market dynamics allowed for it.
Lobbying Away the Guardrails
Having discussed two ways Silicon Valley ventures affect the environment, it’s important to examine existing regulations and how they are also subject to corporate lobbying.
In late 2023, the Biden Administration issued Executive Order 14110, “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence” which created safeguards on the development and usage of AI.
Big Tech felt it was too restrictive and, in early 2024, began lobbying against it.
Silicon Valley identified compute as the bottleneck to AI advancements and focused its efforts on removing the “red tape” that prevented the construction of new data centers and power plants.
Following the release of ChatGPT in 2022, the tech industry saw a 370% increase in AI lobbying spending (300$ million alone in 2025) according to Bloomberg Law, using national security and the manufactured “AI arms race” with China as justification.
The logic was that the bureaucracy of environmental regulations inhibited the US from keeping up with China as they developed better models, with deregulation being the path forward.
OpenAI, Andreessen Horowitz, Google, Microsoft and Amazon all poured money to support this cause, hiring former EPA officials and environmental lawyers to find loopholes in the National Environmental Policy Act (NEPA).
These efforts bore fruit upon securing Trump’s reelection, who signed Executive Order 14179, “Removing Barriers to American Leadership in Artificial Intelligence,” which revoked Biden’s executive order and put out a call for the creation of an “AI Action Plan.”
In response, Google and OpenAI (among others) would submit policy comments calling for regulatory preemption and infrastructure investment.
These points would become codified into law when the White House released “America’s AI Action Plan.”
The plan pushed forward a lot of deregulation that Big Tech had been waiting for. It established categorical exclusions for data centers under NEPA, allowing them to skip multi-year reviews.
Data centers are fast-tracked past the Clean Water Act and Clean Air Act to allow companies to begin construction near wetlands without review.
Lastly, it made federal lands available for data center construction to avoid stricter state-level protections in places like California.
Circumventing these laws allows for rapid data center construction and avoids the costs of building more expensive, but more renewable, energy.
The downstream environmental impacts throw the underlying motivations into sharp relief.
Efforts to make data centers more sustainable primarily serve to greenwash reputations, as the low cost of carbon-intensive energy continues to take precedence over sustainability.
The Techno-Utopian Justification
For Silicon Valley’s role in damaging the environment, sustainability is a recurring talking point in justifying its velocity.
The underpinnings of these arguments are the ideologies of longtermism and techno-utopianism. It’s important to first discuss effective altruism, a philosophy advocating for impartially determining which social causes should be prioritized to help the maximum number of people. In tech, it's frequently used as a justification for acquiring as much money as possible, as billionaires deserve their capital because they can then find the best ways to donate it.
Longtermism is the effective altruist argument that you need to consider what has the highest expected value in the calculus of helping others. Since there are many more people in the future, the math points to mitigating existential risks, such as climate change and AI, for future people.
Techno-utopianism is the idea that rapid advancements in science and technology are key to drastically improving society.
Critically, it views technological growth as inevitable and thus should be accelerated, not slowed. Regulatory control is bound to fail because growth is an inevitability.
As such, it would be better for the U.S. to lead the change rather than other entities such as China. In the context of AI, the stated bet is that an advanced enough AI can mitigate climate change.
And while AI has seen use cases in areas like climate modeling and energy optimization, the specifics of how it can achieve these moonshots remain unclear.
However, until the blurry horizon of AGI (artificial general intelligence) is reached, companies should continue funding the development of more advanced models and data centers.
These arguments only serve to obfuscate the fact that market logic is the ultimate determinant of corporate behavior.
Lobbying to water down regulation, building technology designed to fail, relying on child labor and extracting natural resources while claiming to be green for your customers isn’t the fault of individual bad actors, but merely rational economic behavior under capitalism.
Legitimate sustainability efforts have been put forth, as companies have tried to implement greater transparency about their emissions and invest in more sustainable technology.
However, the drivers of these impulses are always what is best for business.
The Case for the Collective
None of this is to say that people are helpless in the face of Silicon Valley’s capital and dreams.
The Right to Repair movement — an ongoing campaign advocating for consumers' ability to repair and modify products they own — is a great example of how coalition building can influence policy.
Planned obsolescence led to lower-quality products that break rapidly, becoming the norm. Companies then prevented consumers from repairing through hardware sabotage and the guise of “copyright infringement” under the Digital Millennium Copyright Act, according to Vice.
These frustrations would spur a number of strange bedfellows. Farmers were frustrated that they could not repair their own tractors without the expense of “authorized” technicians.
Tech hobbyists and repair shop owners were similarly frustrated by being barred from repairing electronics and environmentalists applied pressure due to the e-waste fallout, according to New York Magazine.
Organizing at the local level involved community events that spread knowledge about repairs and helped fix broken items.
The Repair Association and the US Public Interest Research Group led the national charge by advocating for legislation, with websites like iFixIt helping democratize repair information.
Despite heavy lobbying against right-to-repair laws by giants like John Deere, Apple, and AT&T, policy began to catch up. In 2012, an automotive right-to-repair act was passed in Massachusetts, ensuring automobile manufacturers had to sell the same service materials used in their dealerships directly to consumers.
As a result, major automobile trade organizations signed a Memorandum of Understanding to apply the Massachusetts standards across all 50 states.
Minnesota followed up with a right-to-repair law in 2023 covering home appliances, and in 2024, Oregon passed a law banning “parts pairing”, where companies use software to block third-party software.
Consequently, in 2024, Apple began allowing consumers to repair screens and batteries, according to the New York Times, and John Deere announced in 2023 that it was signing a Memorandum of Understanding allowing farmers to repair their own equipment.
A popular lie parroted by industry is that one’s capacity for environmental resistance is limited to one’s consumption.
The whole idea of the “carbon footprint” was popularized to shift responsibility from corporations to personal lifestyle choices. This seeks to mask the fact that collective organizing has continually been a driver of meaningful change.
Boycotts and divestment movements are examples of negative resistance, defined as refusing to participate in the system when participation falls below one’s tolerance threshold.
Advertisements, industry, and government have a very limited ability to force participation. For the individual, intentional consumption is a small but easy starting point.
While the financial impacts on corporations might be negligible, the value in these small choices is to demonstrate how easy it is to avoid participation.
By rehearsing resistance on a small scale, people are clued into civic engagement and learn what they can demand and act on.
This recognition translates into sustained, organized pressure that can shape policy and corporate behavior.
In the past year, bipartisan grassroots movements across the U.S. stymied $64 billion of data center projects, according to Data Center Watch.
The influence of Silicon Valley can feel staggering, and under the status quo, sustainability runs in direct opposition to survival.
Still, while a truly sustainable path would require staggering shifts across culture, policy and the economy, the resources for less extractive technology exists; what remains is adjusting the incentives.