The bubble surrounding artificial intelligence is bound to burst, and it’s crucial that we approach its aftermath with care and responsibility.
Let’s rewind to December 1999. Tech investors were in a state of euphoria, believing that simply having a website and a flashy Super Bowl advertisement were all it took to generate immense wealth overnight. There was a dangerous misconception where spending was equated with genuine growth, and marketing was mistaken for a legitimate business strategy. Not long after, the dot-com boom would come crashing down, leading to a staggering loss of $1.7 trillion in market value and a broader economic fallout of $5 trillion.
However, from this chaos, something extraordinary arose. The internet that emerged after the crash wasn’t marked by reckless speculation but rather by creativity and innovation: the advent of web 2.0, the rise of open-source software, and the birth of groundbreaking platforms like Firefox and Wikipedia. The takeaway here is quite clear: while bubbles can burst, they can pave the way for better alternatives if we make deliberate choices in how we rebuild.
Today, we find ourselves in a similar situation, but this time it revolves around AI technology.
The current AI boom has an unsettling resemblance to the events of the late '90s. In fact, nearly 80% of the gains in the stock market during 2025 are concentrated among just seven major companies: Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla. These corporations are all striving for dominance over the entire AI ecosystem, which encompasses hardware, software, data, energy, and infrastructure. This struggle goes beyond merely capturing market share; it fundamentally shapes how billions of people will learn, create, and perceive our world.
This concentration of power should raise alarms for all of us.
Similar to the dot-com era, the valuations of these companies are soaring without any clear path toward sustainable profitability. The narrative being pushed suggests that AI will take over human jobs, despite the fact that a staggering 95% of corporate AI initiatives fail to transition into production. Rather than focusing on developing tools that benefit the public and enhance human capabilities, much of the industry seems to be generating what Cory Doctorow refers to as "productive residue" — an overwhelming volume of synthetic media, misinformation, and deepfakes.
The issue at hand isn’t AI itself; it stems from the flawed economic principles currently driving its development.
This scenario is not an unavoidable fate. It arises from a prevailing economic model that treats technology as an extractive industry—accumulating data, centralizing power, and externalizing negative impacts. The race in AI innovation is not propelled by groundbreaking advancements, but by the pursuit of dominance, prioritizing profits over human welfare.
Fortunately, there is a viable alternative economic model already in play.
Around the globe, innovative open-source developers and purpose-driven companies are constructing shared infrastructures that promote trustworthy AI. These systems are designed to be transparent, auditable, and adaptable to local contexts, demonstrating that true innovation doesn’t have to rely on monopolistic control of data.
This movement is exemplified by leading companies whose founders are dedicated to creating tools that are both principled and competitive. For instance, Hugging Face operates the most widely used hub for open-source machine learning models and datasets; Flower AI facilitates decentralized, federated learning to counter the dominance of centralized models; and Oumi provides a fully open-source platform for developing and deploying custom AI models on local infrastructures instead of proprietary cloud systems. And there are many more like them.
These ventures are not mere speculative investments; they represent the foundation for a more sustainable and diverse technological ecosystem. We envision this as part of a double-bottom-line economic model for technology, one that values both mission and profit.
Mediocrity is not our destiny.
If history serves as a guide, the current excitement surrounding AI is likely to end in a manner similar to the dot-com boom: with a crash. However, this doesn’t signify the conclusion of progress; rather, it heralds the dawn of a new chapter.
During the last bubble, the Linux stack—the open-source components that now support much of the internet—emerged from the ashes to surpass Windows. Over the past twenty years, such open-source frameworks have generated an astonishing $8.8 trillion in value, with new studies estimating tens of billions of dollars in potential value for startups and businesses willing to transition from closed AI platforms to open-source alternatives.
What kind of value can we unlock today? The possibilities are immense.
When the AI bubble eventually bursts, we will face a pivotal choice. We could choose to reconstruct the same monopolistic framework, or we could seize the opportunity to develop an economy that prioritizes humanity and shared values. This means adopting open models, ensuring transparent governance, and promoting fair participation in the value generated by AI.
Moreover, it requires us to concentrate on what individuals truly desire from technology: privacy, security, agency, and joy. The real promise of AI lies not in its limitless scalability, but in its potential to enrich our lives, making them easier and more creative without infringing upon our rights or dignity.
Such changes are already taking place. As we experiment with privacy-focused, open-source solutions for everyday tools like browsing and email assistants, we observe consistent improvements in their effectiveness.
Imagine a future where individuals and communities have the capability to host small, localized AI models—energy-efficient, privacy-respecting, and tailored to specific needs. A setting where developers collaborate on tool creation instead of competing against each other. An environment where innovation is assessed not by market dominance, but by its contribution to the public good.
This vision is not just an idealistic dream. If we begin now—constructing AI systems that are open, transparent, and founded on collective values—we can ensure that the next technological era enhances human freedom rather than limiting it. Just as the dot-com crash led to the modern web, the next downturn could pave the way for an even better future—if we dare to rethink the economics of innovation.
Ultimately, the decision lies with us. We can allow a select few companies to dictate our future, or we can collectively take ownership of what we create.
