GPT-5.4 Just Beat Humans. Here's What Comes Next.

The AI Revolution: How GPT-5.4 is Changing the Game and What Comes Next

The recent breakthrough of OpenAI’s GPT-5.4 ‘Thinking’ model, which scored 83% on the GDPVal benchmark, marks a significant milestone in the artificial intelligence landscape. This achievement means that GPT-5.4 can match or even surpass human experts in real economic tasks, making it a potential game-changer for various industries. As we delve into the implications of this development, it’s essential to understand that the real story isn’t just about the model itself, but about the underlying compute war that’s driving innovation and the potential power crisis that could derail progress by 2028.

The Compute War: The Driving Force Behind AI Innovation

The reason GPT-5.4 achieved human-level performance isn’t due to a clever new algorithm, but rather brute force compute. Since 2012, the amount of computing power used to train the largest AI models has grown by more than 300,000 times. To put this into perspective, Moore’s Law would have only delivered a 7x increase in the same period. This exponential growth is a testament to the fact that AI is growing 40,000 times faster than the chip industry that powers it. The compute used per AI model is doubling every 3.4 months, resulting in a 10x increase every single year, compounding. As Elon Musk put it, applying 10x more compute to a language model effectively doubles its intelligence, and the scaling laws are holding firm.

Some key points to consider:

• The compute war is driving innovation in AI, with companies competing to build larger and more powerful models.
• The growth of AI is outpacing the chip industry, leading to a potential shortage of computing power.
• Brute force compute is currently the primary driver of AI innovation, but this may not be sustainable in the long term.

The Power Crisis: A Potential Roadblock to AI Progress

As AI companies continue to push the boundaries of compute power, they’re now spending more than 80% of their total capital on compute resources. This includes chips, electricity, and cooling systems. While this investment is driving innovation, it’s also leading to a potential power crisis. Morgan Stanley recently released a report warning of a U.S. power shortfall of 9 to 18 gigawatts by 2028, which could lead to a 12 to 25% deficit in the power required to run the AI revolution. This is a critical issue that needs to be addressed, as the lights may literally go out on the AI revolution.

Some potential consequences of the power crisis:

• AI innovation may slow down or even come to a halt due to a lack of computing power.
• The cost of AI development may increase significantly, making it less accessible to smaller companies and individuals.
• The environmental impact of AI development may become a significant concern, as the demand for power continues to grow.

The 15-15-15 Dynamic: A New Era in Data Center Economics

Morgan Stanley has identified a new dynamic in data center economics, which they call the ‘15-15-15’ dynamic. This refers to 15-year leases, at 15% yields, generating 15 dollars per watt in net value creation. This is a significant shift in the way data centers are being valued, and it’s turning the tech industry into a real estate gold mine with software margins. Companies that are building data centers and controlling power, data centers, and chip supply are the ones that will win in the next decade.

Some key points to consider:

• The 15-15-15 dynamic is changing the way data centers are valued and invested in.
• Companies that control power, data centers, and chip supply will have a significant advantage in the AI landscape.
• The tech industry is becoming increasingly intertwined with the real estate industry, as data centers become a critical component of AI development.

The Future of Work: How AI Will Change the Job Market

The AI revolution will have a significant impact on the job market, with many tasks being automated and jobs being replaced. Morgan Stanley predicts that AI will be a deflationary force, enabling companies with just 1 to 5 employees to outcompete large incumbents. This means that small startups with the right AI tools can beat large companies, making it a level playing field. However, this also means that many jobs will be lost, and workers will need to adapt to new technologies and workflows.

Some potential consequences of AI on the job market:

• Many tasks will be automated, leading to job loss and displacement.
• New jobs will be created in fields such as AI development, deployment, and maintenance.
• Workers will need to develop new skills to remain relevant in an AI-driven economy.

The Contrarian Take: AI Will Not Create More Jobs

While many people believe that AI will create new job opportunities, Morgan Stanley takes a contrarian view. They predict that AI will be a deflationary force, enabling companies to produce more with less labor. This means that AI will not create more jobs, but rather replace existing ones. This is a critical issue that needs to be addressed, as the impact of AI on employment will be significant.

Some potential consequences of AI on employment:

• AI may lead to significant job loss, particularly in industries that are heavily reliant on manual labor.
• The cost of AI development and deployment may be prohibitively expensive for small companies and individuals.
• The benefits of AI may be concentrated among a small elite, leading to increased income inequality.

The Recursive Self-Improvement Loop: A Potential Game-Changer

xAI co-founder Jimmy Ba has suggested that recursive self-improvement loops in AI could emerge as early as the first half of 2027. This means that AI systems will be able to improve themselves without human intervention, leading to an exponential increase in intelligence. This is a potential game-changer, as it could lead to significant breakthroughs in AI research and development.

Some potential consequences of recursive self-improvement loops:

• AI systems may become significantly more intelligent and capable, leading to breakthroughs in fields such as medicine and finance.
• The development of AI may become more rapid and autonomous, leading to significant changes in the tech industry.
• The risks associated with AI may increase, as AI systems become more powerful and autonomous.

What This Means for You

So, what does this mean for you? First, every job that involves processing information, writing reports, analyzing data, or making routine decisions is on the clock. Not in 10 years, but in the next 24 months. Second, the value isn’t in using AI tools, but in building things that compound on top of them. Third, geography matters again, as companies near cheap power and water are about to become the new tech giants.

Some actionable advice:

• Learn the tools: Not just the basics, but the advanced workflows that can help you build and deploy AI systems.
• Watch the compute leaders: The companies controlling power, data centers, and chip supply are the ones that will win in the next decade.
• Start building something: Anything, right now. The window is closing faster than anyone is admitting, and the compute curve is getting steeper.

The Biggest Mistake People Are Making in 2026

The biggest mistake people are making in 2026 is treating AI like it’s a normal technology. Cars came along, society adapted over 50 years. The internet came along, society adapted over 30. AI is moving faster than both, and the compute curve says it’s going to keep accelerating until something physical stops it. That something is probably the power grid. But we have at least 18 months before that constraint hits. That’s your window.

Some key points to consider:

• AI is not a normal technology, and it will have a significant impact on society and the economy.
• The compute curve is getting steeper, and AI is moving faster than any other technology.
• The power grid is a potential constraint on AI development, but we have at least 18 months before that constraint hits.

Conclusion

In conclusion, the AI revolution is here, and it’s changing the game. GPT-5.4 is just the beginning, and the compute war is driving innovation. However, the power crisis is a potential roadblock, and the 15-15-15 dynamic is changing the way data centers are valued. The future of work will be significantly impacted, and AI will not create more jobs. The recursive self-improvement loop is a potential game-changer, and it’s essential to learn the tools, watch the compute leaders, and start building something. The biggest mistake people are making in 2026 is treating AI like it’s a normal technology. Don’t make that mistake. Take 30 minutes today and write down three things: what does your job look like if 80% of it can be done by AI in 24 months, what would you build if you had 10 employees worth of AI agents working for you for free, and where will electricity be cheapest in 5 years. Then start moving. The window is closing faster than anyone is admitting.

This article is part of our tech series. Subscribe to our YouTube channel for video versions of our content.