Daniil and David Liberman

On December 1-2, join us at the first AI Future Forum held in Dubai, where we'll be among the speakers. It's a great opportunity to dive deep into the latest AI trends and meet industry leaders

Use code Gonka for 10% off your ticket: ai-future.com/?utm_source=sp…

When we were creating Gonka, we believed that the only way to build a decentralized protocol, where independent minds and open-source contributors would gather and form the backbone of the community, was to launch it without any legal entity behind it. There was no company or organization to decide what comes next and how Gonka should evolve. Every major decision was made by the community, by those who contribute compute capacity to the network

As Gonka grew larger and more visible, it became more and more obvious that, after all, in some cases it needed a legal entity. Not to control the network, but to represent it in conversations with partners and to streamline many processes around the protocol

Today, the Gonka Foundation was presented to the community. It's an independent non-profit organization that will push the protocol further, provide proper oversight and act as its steward

We don't have any control over the Foundation, and we still believe in the power of decentralization. The next step will be a set of governance votes where the community will decide whether it trusts the Foundation with the right to act on behalf of Gonka

What do you do if you don’t have enough servers?

Recently, DeepSeek released such a successful update to its v4 Flash model that the company nearly ran out of server capacity. Smart enough, fast enough, and cheap enough, the model became so popular that DeepSeek was forced to raise prices substantially in order to curb demand

Kimi found itself in a similar situation when, following the release of K3, it had to temporarily restrict new user registrations

No one is immune to compute shortages caused by a surge in users, not even centralized companies with virtually unlimited resources. Anthropic, for example, came up with “creative” ways to handle the load, from reducing the reasoning level of its models to banning the use of subscriptions in agents such as OpenClaw

But there is one important difference between OpenAI or Anthropic and DeepSeek or Kimi: the latter still release their model weights. That means that where their own servers fall short, dozens of alternative providers can step in neoclouds, coding-agent developers, and decentralized networks. Through market mechanisms, demand is spread across these providers instead of running up against the limits of a single centralized player

Incidentally, DeepSeek v4 Flash also became available on Gonka last week. In just a couple of days, the model became the most in-demand among users, and the network is setting records for the number of tokens generated

Try DeepSeek and other models on Gonka today. Pick any broker from the list and send your first request

WebX was the first major test of an idea: can a community independently bring a decentralized AI protocol onto the global stage? Now we know the answer. It can.

medium.com/product-scienc…

Sat down with @DanielNenni to unpack why Proof of Work incentives are the key to building massive, decentralized infrastructure, how permissionless networks rival centralized clouds in latency, and the incoming wave of hyper-efficient AI ASICs.
Tune in to hear how the open market is solving the hyperscaler bottleneck! 👇

semiwiki.com/podcast/podcas…