Amazon Web Services (AWS) and Microsoft have been pulling back from AI data center investment, suggesting problems with the centralized model. Analysts are taking this latest development to reiterate why decentralized blockchain-based infrastructure could be the solution.
Kai Wawrzinek, co-founder of Impossible Cloud Network, discussed these looming questions in an exclusive interview with BeInCrypto.
AI Data Centers Hit a Wall
A few months ago, AI seemed like one of the global tech industry’s most promising sectors. However, with firms like AWS and Microsoft announcing pauses in AI data center construction, the picture looks very different. What happened? What does the future of AI look like? Kai Wawrzinek described the situation as it stands today:
“News that AWS is joining Microsoft in pulling out of new data centers when demand for AI is growing exponentially is testament to the enormous inefficiency this model presents for scaling the global internet. Microsoft and AWS may be coming to realize that centralized infrastructure models simply can’t adapt fast enough,” Wawrzinek claimed.
AWS and Microsoft aren’t the only companies facing these problems. Although Meta publicly claimed it would spend hundreds of billions on AI infrastructure and data centers, it asked competitors for funding less than three months later.
OpenAI, too, has been rocked by the sheer cost of operating ChatGPT; Sam Altman tacitly admitted that its research may never be profitable.
Wawrzinek sees a clear solution – abandon the centralized model altogether and focus on DeFAI. Although these industry leaders accumulated billions in capex and pioneered LLM development, the entire strategy can be self-defeating.
For example, US AI data center construction is swamping electrical engineers with work to an unprecedented degree. With so many professionals focusing on the centers themselves, it’s creating a bottleneck for skilled labor.
This harms renewable energy projects and the electrical grid, ironically harming the data centers’ functionality.
“The AI era needs infrastructure that can match its speed and scale, and decentralized systems are the only models built for that future. In contrast, a decentralized, market-driven approach solves this problem: capacity can be deployed more efficiently where and when it’s needed without waiting years for centralized megaprojects,” Wawrzinek added.
Can DeFAI Handle the Challenges?
Compared to the centralized data center model, DeFAI has increased AI compute accessibility. Blockchain-enabled economic incentives can accelerate deployment speed, enhance scalability, and optimize resource allocation without massive upfront capital.
These decentralized systems, in short, have more agility than their competitors.
Blockchain-based AI companies have been able to leverage significant compute capacity without centralized data centers. For example, the DePIN firm Aethir has made great strides with its GPU-as-a-service model.
Other firms like 0G Labs have proved that decentralized AI development isn’t just theoretically feasible; it’s profitable and necessary for the ecosystem.
If this all seems far-fetched or utopian, it’s important to remember AI’s “black swan” event – DeepSeek.
China’s market-moving genAI model proved to the entire world that AI firms can make state-of-the-art LLMs at a fraction of the hardware cost. So, the AI industry may need to rethink the data center model altogether if this one developer proved so successful.
Although skeptics have wondered whether decentralized AI can compete with data centers, the reality is that centralization can have its own inefficiencies.
“The future of AI infrastructure lies in open, permissionless networks, where supply meets demand dynamically and globally, not through outdated hyperscaler models that are struggling to keep up,” Wawrzinek finished.
So far, centralized AI firms have accumulated billions in venture capital investment, but their ability to innovate is hitting a brick wall. We may need a better model to create the best possible outcomes.
