AI companies are building huge natural gas plants to power data centers. What could go wrong?

The News

Major AI companies, including Meta, Microsoft, and Google, are aggressively investing in new natural gas power plants to support the escalating energy demands of their artificial intelligence data centers [1]. This shift marks a significant departure from previous sustainability pledges and raises concerns about the environmental impact and long-term viability of AI infrastructure [1]. Meta’s upcoming Hyperion data center, for example, is slated to be powered by ten new natural gas plants, a scale that has been compared to the energy needs of the entire state of South Dakota [2]. Similarly, a new Google-funded data center is projected to emit millions of tons of emissions annually through its associated natural gas plant [4]. The trend, while seemingly localized to North America at present, signals a broader industry response to the insatiable power requirements of increasingly complex AI models and the massive datasets they consume [1]. The announcements, primarily occurring within the past week, have been met with criticism from environmental groups and renewed scrutiny of the AI industry's commitment to carbon neutrality [1].

The Context

The current surge in natural gas plant construction is directly linked to the exponential growth in AI model size and the corresponding increase in computational demands [1]. Training and running large language models (LLMs) like Gemini and Llama 3 requires immense processing power, which translates into significant energy consumption. A data center, as defined by Wikipedia, is a facility used to house computer systems and associated components, such as telecommunications and storage systems. These facilities are critical infrastructure for the storage and processing of information, supporting everything from the global financial system to cloud services and machine learning. The architecture of modern AI data centers is characterized by high-density computing racks, advanced cooling systems, and redundant power infrastructure – all of which contribute to substantial energy footprints [1].

The decision to build dedicated natural gas plants, rather than relying solely on grid power or renewable energy sources, stems from a combination of factors. Grid capacity, particularly in regions targeted for data center expansion, is often insufficient to meet the rapidly increasing demand [1]. While renewable energy sources are desirable, their intermittent nature necessitates backup power solutions, and natural gas plants offer a readily available and relatively controllable alternative [1]. Furthermore, the latency requirements of AI workloads – the need for near-instantaneous processing – often preclude the use of geographically dispersed renewable energy sources due to transmission delays [1]. The speed of AI development, particularly the race to deploy ever-larger models, has created a situation where immediate power solutions are prioritized over long-term sustainability goals [1]. The scale of Meta’s investment is particularly noteworthy; powering a facility equivalent to South Dakota’s energy consumption highlights the sheer magnitude of the power demands [2]. This contrasts with earlier, more aspirational goals of carbon neutrality within the AI sector, now appearing increasingly difficult to achieve with current infrastructure trends [1]. Interestingly, SpaceX recently filed an application with the US Federal Communications Commission to launch up to one million data centers into Earth’s orbit [3], suggesting a potential long-term solution to power constraints, albeit one fraught with significant technological and regulatory hurdles.

Why It Matters

The reliance on natural gas plants for AI data centers has cascading impacts across multiple sectors. For developers and engineers, the situation introduces a layer of technical friction. The lack of consistent, renewable power sources can complicate the design and optimization of AI systems, potentially limiting the efficiency of algorithms and the deployment of energy-saving techniques [1]. Furthermore, the environmental concerns surrounding natural gas plants may lead to increased regulatory scrutiny and potential carbon taxes, adding to the operational costs of AI development [1]. For enterprise and startups, the increased energy costs associated with these plants represent a significant financial burden, potentially hindering innovation and widening the gap between large, well-funded AI players and smaller competitors [1]. Large corporations like Google and Meta can absorb these costs more easily than smaller companies, creating an uneven playing field [1].

The winners in this scenario are primarily the natural gas industry and the companies that provide the infrastructure for building and operating these plants [1]. Conversely, renewable energy companies and those advocating for sustainable AI practices are facing a setback, as the industry prioritizes immediate power needs over long-term environmental goals [1]. The situation also creates a potential conflict for companies that publicly promote sustainability while simultaneously investing in fossil fuel infrastructure [1]. The emissions from these natural gas plants contribute significantly to greenhouse gas emissions, exacerbating climate change and potentially triggering stricter environmental regulations that could further impact the AI industry's operations [4]. The long-term implications extend beyond the immediate financial and environmental costs, potentially impacting public perception of AI and hindering its widespread adoption [1].

The Bigger Picture

The current trend of building natural gas plants to power AI data centers represents a significant deviation from the industry’s previously stated commitment to sustainability [1]. While companies like Microsoft and Google have publicly pledged to achieve carbon neutrality, their actions are increasingly at odds with these pronouncements [1]. This shift is occurring amidst a broader debate about the environmental impact of AI, with critics arguing that the industry’s relentless pursuit of greater computing power is unsustainable [1]. Competitors are facing similar pressures, and the race to deploy ever-larger AI models is driving a prioritization of immediate power solutions over long-term sustainability considerations [1]. The move towards natural gas plants also highlights the limitations of relying solely on grid power and the challenges of integrating renewable energy sources into AI infrastructure [1]. The SpaceX proposal to launch data centers into orbit [3], while currently speculative, underscores the desperation to find solutions to the escalating power demands of AI [3]. This suggests that the industry is actively exploring radical alternatives, but these solutions remain years, if not decades, away from widespread implementation [3]. The next 12-18 months are likely to see continued pressure on the AI industry to address its energy consumption, with increased scrutiny from regulators, investors, and the public [1].

Daily Neural Digest Analysis

The mainstream media’s coverage of this issue tends to focus on the immediate financial and logistical aspects of building natural gas plants [1]. However, the deeper risk lies in the erosion of trust and the potential for a backlash against AI development. The optics of companies simultaneously touting the benefits of AI while contributing to environmental degradation are damaging, and this could ultimately stifle innovation. The reliance on natural gas represents a short-sighted solution to a long-term problem, and the industry is essentially kicking the can down the road. The SpaceX proposal, while ambitious, highlights the desperation to find alternative power sources, and the fact that such a radical solution is even being considered underscores the severity of the situation [3]. The question remains: will the AI industry proactively address its energy consumption and embrace truly sustainable solutions, or will it continue down a path that risks undermining the very benefits it promises?

---

References

[1] Editorial_board — Original article — https://techcrunch.com/2026/04/03/ai-companies-are-building-huge-natural-gas-plants-to-power-data-centers-what-could-go-wrong/

[2] TechCrunch — Meta’s natural gas binge could power South Dakota — https://techcrunch.com/2026/04/01/metas-natural-gas-binge-could-power-south-dakota/

[3] MIT Tech Review — Four things we’d need to put data centers in space — https://www.technologyreview.com/2026/04/03/1135073/four-things-wed-need-to-put-data-centers-in-space/

[4] Wired — A New Google-Funded Data Center Will Be Powered by a Massive Gas Plant — https://www.wired.com/story/a-new-google-funded-data-center-will-be-powered-by-a-massive-gas-plant/