The AI Kamikaze: Inside the Tech Bubble Sidelining the Climate Crisis

Nowadays, the "AI" label is slapped onto everything, but true artificial intelligence is actually split into different levels.

It starts with Narrow AI, which acts like a brilliant but hyper-specialized expert. It can beat world-champion gamers or detect cancer better than doctors, but it is completely helpless outside its specific box. If you change its environment by even a centimeter, it breaks because it doesn't actually "think"—it just solves a single mathematical puzzle.

That brings us to our current reality: Large Language Models (LLMs). This is the more adaptable technology that everyone today considers to be the first glimpse of true, advanced artificial intelligence.

In simple terms, think of today's AI as a collection of specialized tools. You have an AI that is a calculator, an AI that acts as a translator, and an AI that plays chess. They are amazing at their one specific job, but they cannot do anything else. A chess AI cannot translate a book, and a translator AI cannot solve a math puzzle.

AGI, or Artificial General Intelligence, is completely different. Instead of being a single tool, AGI is like a human brain. If you give a human a brand-new task they have never seen before, they don't crash. They use their common sense, look at the situation, figure out the rules, and learn how to do it.

AGI is a computer program that can do exactly that. It can think, learn, and adapt just like a person. It can teach itself how to paint a picture, write code for an app, cook a new recipe, and solve a complex business problem, all using the same digital mind.

The reality is that AGI is far from being implemented. Even with current AI use, there is not enough electricity and infrastructure to support it.

Advancements in AI boomed between 2023 and 2025. November 2025 marked a massive turning point for AI policy, defined by a coordinated, global shift toward deregulation specifically designed to speed up development and help tech companies innovate faster. Governments realized that the tight regulations they drafted in 2023 and 2024 were slowing down progress and forcing businesses to stumble over red tape.

However, despite the aggressive inflow of money into AI, removing policies will not necessarily bring about AGI any sooner. In this rushed and overhyped state, the technology may cause more harm than benefit. While AI is a great tool, this massive over-pushing and its skyrocketing energy demand are leading to severe pollution. Furthermore, the lack of strict restrictions creates serious risks, making it much easier for AI to be used for dangerous activities.

The catch is that, according to research data, we hit the top of the curve in 2025, where progress ground to a total halt. This is what they call the 'tragedy of the final mile'—the reality that going from zero to 90% is easy because nearly all of humanity’s data is readily available to get you there. But achieving that final 1% isn't just ten times harder and more expensive; it’s a thousand times harder than everything else combined. And this isn't just true for AI; it applies to absolutely anything in life.

It is incredibly difficult and risky to admit at this stage that AI stocks and investments are heavily overpriced. We appear to be heading toward a dot-com-style bubble reminiscent of the late 1990s. Despite severely overvalued stocks and the complete absence of a market correction, investors remain overwhelmingly bullish.

The primary driver behind this dangerous euphoria is a massive circular flow of capital. The world's largest tech giants are pouring billions of dollars into building enormous data centers and purchasing high-end microchips, creating a temporary illusion of infinite financial growth. However, look closely at the other side of the equation. While infrastructure providers are reporting record-breaking profits, the companies actually developing the software are hemorrhaging cash. Operating costs and energy demands are skyrocketing, yet these generative models have failed to present a clear, sustainable roadmap to true profitability.

We are seeing a profound disconnect between corporate valuations and financial reality. When millions of users type prompts into a chatbot, the computing power required costs the parent companies millions of dollars a day. If these AI companies cannot find a way to monetize their products effectively before their massive cash reserves run dry, the market will face a reality check. Just like the 1990s internet boom, the technology itself will ultimately change the world, but the first wave of overhyped businesses funding it will likely face a devastating collapse.

Before the AI boom, climate change was the dominant global headline. Today, it seems to have completely evaporated from public discourse. Consider an important fact: even before AI took off, electricity generation was already the second-largest contributor to global greenhouse gases. Yet, while everyday citizens are busy tracking their personal carbon footprints and switching to electric vehicles—a frustrating paradox when those cars still rely on fossil-fuel-powered grids—the world's largest market players are quietly making climate management significantly worse. This corporate environmental damage is barely even mentioned in the news anymore; it has simply disappeared.

While AI data centers require so much energy that is not even manageable at the moment. Instead of decreasing greenhouse gases, current AI infrastructure development is threatening to do the exact opposite. A prime example is a massive new project in northwestern Utah, where county commissioners recently approved a development agreement for a staggering 40,000-acre hyperscale AI data center campus. The approval went through despite fierce backlash and thousands of formal complaints from local residents who fear the project will destroy the local environment and drive up energy costs.

The $100 billion mega-project, dubbed the Stratos Project, is spearheaded by O’Leary Digital, an infrastructure firm helmed by venture capitalist Kevin O’Leary. At full capacity, this massive hyperscale campus will require an unprecedented 9 gigawatts of power—a figure that is more than double the average electricity consumption of the entire state of Utah.

While proponents promise thousands of local jobs, independent analyses suggest a fraction of those numbers is far more realistic over its multi-decade rollout. Meanwhile, environmentalists and scientists are sounding the alarm over the sheer scale of the facility. Critics warn that building a natural-gas-reliant monster of this magnitude could fundamentally imperil the already fragile Great Salt Lake ecosystem. Furthermore, the immense waste heat generated by its cooling fans threatens to alter the local climate directly, potentially spiking local daytime temperatures by up to five degrees, and nighttime temperatures by a staggering 12 to 28 degrees.

Somehow, over the past three years, the world went completely kamikaze with AI. Climate change—once universally recognized as our greatest and most urgent global risk—has been entirely sidelined; we can barely find news or meaningful policy updates on it anymore. To break this destructive cycle, we desperately need a major market correction, a return to normalization, and a significant slowdown in this frantic AI spending. The sheer speed and relentless bullishness of today's markets mirror the reckless euphoria of the late-1990s dot-com bubble and the systemic blind spots that led to the 2008 financial crisis. This article focuses solely on this technological and economic frenzy, leaving aside the ongoing global wars, which I have thoroughly analyzed in my previous posts.