Alexander Dilkow

RAG is broken and nobody's talking about it. Stanford researchers exposed the fatal flaw killing every "AI that reads your docs" product in existence. It’s called "Semantic Collapse," and it happens the second your knowledge base hits critical mass. If you've noticed your AI getting "dumber" as you add more data, this is exactly why. Right now, companies are dumping thousands of documents into their AI, thinking it’s getting smarter. When you add a document to RAG, it converts it into a high-dimensional vector. Under 10,000 documents, this works perfectly. Similar concepts cluster together. But past 10,000 documents, the space fills up. The clusters overlap. The distances compress. Everything starts to look "relevant." It is a mathematical law called the Curse of Dimensionality. In a 1000-dimensional space, 99.9% of your data lives on the outer edge. All points become equidistant from each other. That perfect, relevant document you are looking for now has the exact same mathematical similarity as 50 completely irrelevant ones. The Stanford findings are brutal: At 50,000 documents, precision drops by 87%. Semantic search actually becomes worse than old-school keyword search. Adding more context doesn’t fix the AI. It makes the hallucinations worse. Your "nearest neighbor" search isn't finding the best answer anymore. It's finding everyone. We thought RAG solved hallucinations. It didn't. It just hid them behind math.

Yes!

@cb_doge If wattage is currency, Habitat is the exchange. No more stranded energy goes to waste. It becomes productive power.

Our synergy solves multiple problems at once! Habitat has started discussions with enterprise compute customers and I welcome you to join us!

@it_do_just @scottmelker As @elonmusk said, you can’t fake energy. You can’t fake network routing or compute output either. Every move we make connects energy, compute and liquidity into one system designed to print real compounding value.

Crypto has a huge problem – and it’s one we almost never talk about honestly. How do founders make hundreds of millions (or billions) from tokens they created before their products prove real utility or adoption? In traditional markets, value follows performance: founders get rich because their company builds something people use, generates profit, and grows over time. In crypto, it’s often reversed. Tokens launch first, valuations skyrocket on speculation, and founders cash out long before the product is battle-tested or the network achieves meaningful usage. The result is a strange inversion of incentives – enormous wealth built on potential rather than proof. And because tokens trade globally, that value is extracted from retail participants who are effectively subsidizing the experiment. I’m not saying founders shouldn’t be rewarded for innovation. Many are building genuinely transformative tech. But when billion-dollar fortunes are realized before success is earned, it raises a fundamental question: If the future value is still undetermined, why are the rewards already realized? That disconnect is what keeps so many people skeptical of token models. Until the industry ties founder upside more closely to actual adoption and network usage, the “speculation over substance” narrative will remain deserved.

The future of AI infrastructure won’t be ruled by monopolies like AWS, Azure, or Google Cloud. It will be fair, sovereign, decentralized, and transparent. This future has a name - Habitat. We’re not building another cloud. We’re building freedom for AI.

The next mission-critical milestone for us is forging partnerships with top EPCs. These partners will deliver Habitat-Ready PV Parks - Next-generation solar infrastructures where Modular Data Center connectivity and siting are built into the blueprint from day one. Here is me signing a MoU with the managing director of EPC Sunnovate.

We are in Singapore today at Solana APEX meeting with DePIN, AI, and infrastructure projects. We’re excited to connect with other founders and talk about the future of infrastructure on Solana. Come say hi! DM's open. luma.com/apex-singapore…

Global solar installs surged 64% in H1 2025 But growth without efficiency = waste. Habitat’s mission is simple: turn waste into value. By solving curtailment, we capture stranded solar energy and convert it into compute. We’re also working on establishing a service where new solar parks are built Habitat-native, designed so every watt is monetized. Grid first, with any curtailed output instantly absorbed by colocated data centers and mining units. 100% efficiency, 2–3x faster break-even. This is how Habitat eliminates waste today and prevents it tomorrow, all through the Habitat token economy. ember-energy.org/latest-updates…

What goes into an industrial-scale solar park? Here’s the full breakdown of Trakia’s hardware: trakiasolar.com

Trakia Solar is owned by one of our founding members. One of the main reasons we prioritized infrastructure first in our project was to clearly identify the opportunity created by curtailment and to fully understand the revenue mechanism, with an experienced operator on our team from the very beginning. Another key benefit of bringing a park owner in early was access to their subject matter expertise and their network of operators at this scale, both on the supply chain side and in business operations, revenue forecasting and future investment discussions. I'm happy to answer any other questions you might have!

@HabitatProtocol You own the solar parks or you partner with them ?

Almost all of the world’s electrical + data center infra was built for Web2. Web3 just rents it. Habitat’s mission is to flip that: build crypto-native infra from the ground up, starting with stranded solar → renewable-powered AI compute. We’ve been hard at work: • Laying the groundwork for first DePIN partnerships • Networking ahead of Token2049 Singapore • Onboarding more solar parks - Node 0 is just the beginning • Building our multistaking protocol and marketplace Our projections show ROI 2–3x faster for solar parks adopting our model. We’re already negotiating access to 1.5 GWh of renewable energy, the foundation of a new internet-native capital market for tokenized compute.

The Habitat Flywheel. An Internet Capital Market where curtailed renewable energy is turned into compute for AI, then tokenized and traded onchain.

New and innovative green power solutions come online every day, creating energy, life and progress. zmescience.com/ecology/japan-…

AI is becoming the grid. We’re building its natural habitat.

Everyone thinks AI is bottlenecked by chips. The truth is bigger. AI is bottlenecked by energy.

By 2030, AI compute will outpace today’s energy grids. Habitat’s edge: turning curtailed solar into guaranteed AI capacity.