The artificial intelligence revolution is no longer a distant whisper; it’s a roaring gale, rapidly reshaping economic landscapes and, crucially for our sector, driving an unprecedented surge in global energy demand. As an oil and gas financial journalist, the implications of this accelerating AI boom on power grids, commodity markets, and investment strategies are front and center.
A leading voice in this computational maelstrom, Lin Qiao, former Meta engineer and CEO of the $4 billion startup Fireworks AI, paints a vivid picture of this exponential growth. Her company, which provides an inference cloud platform, is now processing an astonishing 15 trillion AI tokens daily. This figure represents a significant leap from the 13 trillion tokens just a few months prior and 10 trillion tokens observed in late 2025. Qiao’s insights confirm what many in the energy industry are beginning to realize: the appetite for AI processing is insatiable and escalating at a dizzying pace.
For context, these “tokens” are the foundational numerical units into which AI models break down words and other inputs for processing. Roughly three-quarters of a word constitutes one token, and their usage forms the basis for pricing AI model consumption at a standard cost per million tokens. The sheer volume of these tokens underscores the immense computational load being exerted across the global digital infrastructure, a load that translates directly into a colossal draw on electrical power.
The Energy Nexus: AI’s Insatiable Power Demand
Qiao’s experience building PyTorch, the open-source framework that fueled the initial wave of modern AI adoption, provided a front-row seat to the technology’s formative years. Back then, purpose-built GPUs for AI were non-existent, and the necessary tooling was nascent. “We had to build everything from the ground up,” she reflected, highlighting the nascent stages of an industry now exploding in scale. The current acceleration, however, is far more dramatic, rapidly embedding AI into every conceivable workflow.
This widespread adoption transcends traditional tech teams. Finance departments are leveraging AI for sophisticated forecasting, legal teams are deploying internal AI tools for efficiency, and even gig economy workers are utilizing generative AI for creative output. Qiao’s observation that her college-age daughter simultaneously uses multiple AI systems – one for generating answers, others for verification – epitomizes how deeply integrated these tools have become into daily life. “Literally every single person is using these tools,” she asserted, underscoring the universal nature of this demand.
Such pervasive usage creates a profound ripple effect throughout the entire technology stack. We’re witnessing tight supplies of GPUs, rising hardware costs, and, critically for our readership, immense strain on power infrastructure. Qiao’s blunt assessment, “The whole system is saturated,” is a stark warning that bottlenecks now extend from semiconductor components all the way to our fundamental energy grids.
Conventional Fuels: Powering the Digital Frontier
This saturation of energy grids represents a significant and often underestimated challenge, but also a monumental opportunity for the oil and gas sector. As AI data centers proliferate and scale, their energy demands are measured not in kilowatts, but in megawatts, often requiring consistent, reliable, and dispatchable power sources that renewables alone cannot yet fully provide without significant backup.
The rapid deployment cycles of new AI models and hardware, with new Nvidia chips arriving every few months and new AI models emerging every few weeks, mean that energy supply must keep pace. While hyperscalers like Amazon, Google, Microsoft, and Oracle offer GPU rentals, the complexity and speed of evolving AI necessitate specialized services like Fireworks AI to optimize performance and manage infrastructure. Similarly, the complexity of meeting this unprecedented energy demand falls squarely on the shoulders of the energy industry.
Natural gas, with its flexibility, lower emissions profile compared to coal, and capacity for rapid deployment of combined-cycle power plants, is poised to become an increasingly critical feedstock for AI data center electricity generation. Its ability to provide continuous, base-load power, and to ramp up quickly to meet fluctuating demand, makes it an ideal partner for the intermittent nature of many renewable energy sources. Oil, while less common for primary grid power, could see increased demand for peaking plants or backup generation in energy-constrained regions where AI infrastructure is rapidly expanding.
Investment Strategies for the AI Energy Boom
For investors in the oil and gas space, the AI boom presents compelling strategic considerations. The exponential growth in token consumption translates directly into an escalating need for electricity, which in turn necessitates significant capital expenditure in new power generation capacity, transmission infrastructure, and, crucially, the upstream and midstream assets that supply the fuels.
Companies involved in natural gas exploration and production (E&P), liquefied natural gas (LNG) export, and midstream infrastructure (pipelines, storage) are likely to see sustained and growing demand. Similarly, power generation companies with portfolios heavily weighted towards natural gas-fired plants could experience enhanced revenues and improved capacity factors. Even oil producers might benefit from overall tightening energy markets and increased industrial demand for various petroleum products related to infrastructure buildout.
The speed at which AI is integrating into industries, from agriculture to manufacturing – a trend Qiao observed during PyTorch’s early days and sees accelerating now – suggests a broad-based, long-term increase in energy consumption. This is not merely a transient tech fad; it is a fundamental shift that will redefine global energy requirements for decades to come.
In essence, the lesson from both the genesis of PyTorch and the current scale of Fireworks AI’s operations is clear: once AI becomes broadly accessible and usable, adoption accelerates dramatically. Based on current token volumes and Qiao’s assessment, this acceleration has only just begun. Savvy investors in the oil and gas sector must recognize that the digital frontier of AI is simultaneously creating an unprecedented demand for reliable, scalable, and increasingly conventional energy resources, positioning our industry at the very heart of this transformative technological era.
