Last year I covered the Network State Conference in Singapore. This year I was remote. You can watch the full replay here. If you want to see past reviews, you can check out my highlights (2025), lessons and experiments (2023-24) posts.
This year’s program covered cryptography, privacy, geopolitics and techno-capitalism, and extended discussion springing from Hirschman’s ideas of exit and voice. Each project also had a booth in the main hall, which made the day feel less like a fixed schedule of talks and more spontaneous for attendees.
Network + State
In 2025 the pitch was exit. This year, Balaji opened with a photo of himself meeting Kazakhstan’s prime minister. He arrived in July planning to stay three days. Within a week Network School had signed MOUs, found its new home, and brought the crypto regulator on board. He said ten other countries have since bid for similar deals.
He called the arrangement “network plus state”: a middle ground between internet communities and nation-states, much as stablecoins sit between fiat and Bitcoin. Kazakhstan’s selling point is space, with about 3,000 times Singapore’s land area and only three times its population.
San Francisco can incubate startups, he argued, but Kazakhstan can incubate startup cities.
Monastery to Embassy
Deputy Prime Minister Zhaslan Madiyev said more than 90% of the country’s 1,300 public services are online, digital IDs have the same legal standing as physical ones, and Alatau City is an 88,000-hectare zone run under English common law.
Network School now occupies a full floor of the government’s AI center in Astana, three floors below an AI-driven government office. Last year, Vitalik described NS as a “monastery.” It has now moved about as far toward his “embassy” end as it could go.
My May 2026 review of Network School is also already out of date. According to the Deputy PM, more than 200 people followed NS to Kazakhstan in under three months, which suggests the community moves with the project.
The broader movement is still in its own incubator phase, so I expect nodes to keep pivoting, relocating and folding. I think the emotional cost is under-acknowledged: During NS’s exit from Malaysia, members had to choose at short notice, between their community and a move to a new country.
Sovereign Money
Privacy and sovereignty over money were pervasive themes through the whole day, starting with a morning of nine Zcash talks. Balaji’s framing was that Bitcoin becomes the transparent reserve asset of Western institutions and Zcash the encrypted cash of individuals.
Josh Swihart said his bank of 30 years debanked him, and that he was later summoned to the White House and told to find a way to backdoor Zcash. Later in the day Balaji argued that because so much Bitcoin now sits with custodians, a modern version of the 1933 gold seizure is plausible, and that it could push the global price up even as holders in the seizing country lose everything.
None of the speakers addressed the tension in the running order: the privacy summit came straight after a minister’s pitch for state partnership, and Kazakhstan is not a liberal democracy.
My read: This is a two-way arbitrage, and a reasonable one. Kazakhstan gets an international network’s talent, capital and attention, and the network gets land, permits and a legal home. The deal stays healthy only while members can still leave, and private money, portable credentials and second passports are what protect that ability.
Fostering enduring communities and cities means leaning towards “wants” and not “shoulds” (voluntary opt-in/out, not coercion).
Token Rate
The topic of the week was AI pointed at classical cryptography. Haseeb noted that most cryptography is trusted only because human experts tried and failed to break it, so a month of concentrated AI effort could expose holes. Rather than Justin Drake‘s “bunker mode”, he argued that chains should agree in advance on a recovery mode that falls back to slow, hash-based cryptography. Vitalik expects public-key cryptography to survive with “a couple of extra Y2Ks.”
Every chain needs a defense budget, meaning some of its own tokens spent on AI that keeps auditing its code and math, and Balaji suggested measuring it like hash rate, as a “token rate.”
The result is a circular economy in which AI agents buy crypto to transact and protocols buy AI to stay secure.
Tokens fund tokens to secure tokens.
Haseeb added that AI attackers tend to converge on the same ideas, so whoever spends most on defense will probably find the bugs first. My takeaway: defenders should run a mix of frontier models for cybersecurity evals, since that covers more ground than any one attacker’s model.
In the end, blockchains become more resilient, secure, and lindy.
Founder Mode
The Vitalik and Balaji session ended on succession. Asked how much the Astana node depends on him, Balaji described a goal he calls “minimum necessary Balaji”: to make himself less essential by developing a hundred startup society leaders. Vitalik’s measure of progress for Ethereum is the community defending its values without waiting for him to speak.
The only thing worse than having a leader is having no leaders. (Balaji)
Decentralization… it’s about going from one leader to many. (Vitalik)
This is the core question for the whole sector: who leads, and why should anyone follow? Balaji’s answer leans center-right (the founder decides because they built it), and Vitalik’s leans center-left (transparent, low-bureaucracy structures that surface leaders from within).
In my opinion, either way, whoever ends up in charge still has to run the thing akin to something like “founder mode”. I would consider the example of someone like Satya Nadella, who didn’t exactly found Microsoft, but rose through the organization and was selected internally to lead the company, in the style of a founder.
By extension, every startup society needs a way to select, filter, and retain good leaders, with incentives aligned to the community. Tech communities need well-rounded leadership that pairs high IQ with high EQ.
A Hundred-Year Node
Pratik Suthar of the RSS described a self-funded organization nearly a century old, with about 10 million members and 89,000 daily gatherings across India. There is no formal membership, so you join by showing up, and members are expected to produce more than they consume. He called it an open-source social impact model that is “designed to be deleted.”
The RSS is also a Hindu nationalist movement with a contested history (it is widely described as the ideological parent of India’s ruling BJP), and I’m not holding it up as a model. What interests me is what its shared religious and national axioms buy: trust, reciprocity and a common story already exist before anyone arrives at an in-person popup or permanent node.
Network state experiments and tech communities aim to recruit people from diverse backgrounds, so the open question is how they build that level of trust without an inherited foundation, in a system this globally decentralized.
Pop-ups Go Permanent
Edge City reported 12,500 attendees across ten events and is raising its first fund ahead of a new flagship in Goa. Mu brought 2,000 people to Shanghai, arranged more than 100 visas with local government and offered firewall-free Wi-Fi, and it is now building a permanent hub there. Noma Collective, with 60,000 members, opened a permanent base in Freeport, Bahamas, and plans hubs in Próspera and Paraguay by the end of 2027.
The more interesting shift was toward families. Ridgerunner builds neighborhoods in Tennessee and Kentucky for households with roots and going concerns, on the logic that “those who come easily leave easily,” and it reports about 100 households so far.
Last Year’s Promises
Ipê City held its 2026 pop-up as promised, with 600+ participants, double last year.
Ârc went from a community inside NS to its own campus in Mirissa, Sri Lanka, with a town planned next.
Edge City‘s Esmeralda and Gelephu projects are still in progress, and India is the new flagship.
Collective visas, which I predicted last year, have partial evidence: Kazakhstan says it built its nomad visa regime with NS’s help, and Mu secured visas for its community.
Not mentioned this year: Infinita, Próspera Montana, Zuitzerland, Liberland, Forma.
Exit and Voice
Mike Maples summarized Hirschman’s idea in one line: “exit is what gives voice its power.” He also suggested someone write Built to Die as the honest sequel to Built to Last. Eric Weinstein offered the best metaphor of the day: near the poles, magnetic north and true north can point in opposite directions, and many Western institutions have drifted the same way, until what they claim to pursue and what they actually pursue are opposites.
Curtis Yarvin argued that America can be rebooted, citing how quickly postwar Germany was reprogrammed, and said that “human beings are natural power followers” more than we like to admit.
He and Balaji closed on a question that sums up the day: is America in the late Republic stage, or the late Empire stage?
Niall Ferguson argued the skeptic’s case, and the day was better for it. In his view, empire is back, the US and China are the only two AI empires, middle powers are deluding themselves, and talent still flows overwhelmingly to America. He also co-founded UATX on the premise that Harvard can’t be fixed, which puts him in the same business as Balaji: building new universities.
The United States is the network state. It’s the network states plural. Fifty of them. (Niall Ferguson)
Passport as Portfolio
The identity block made a simple case: states treat you as a subject unless you have options. Henley & Partners said US citizens are now more than 30% of its clients, and CitizenX quoted Balaji back to the room: cash is king, but passport is queen.
The $2B School
Balaji saved the biggest news for last. Polychain is leading a round in Network School at a $2B valuation, with a16z and Coinbase participating, and Kazakhstan’s education ministry has signed an MOU to make NS a degree-granting university.
Students graduate by exam rather than by years of attendance, meet in residential cohorts about one week per quarter, receive diplomas on-chain, and have a job-placement deal with Kazakhtelecom.
Alatau City is the physical side of the bet: a greenfield site of 880 km² targeting 1.8 million residents by 2050, “tokenized by default,” with operations starting in 2027.
Balaji’s twist is value capture. If residents hold a stake in the city instead of a house, they have a reason to support new building rather than block it. Aligned YIMBYism!
Where I Land
Some things I’m tracking looking ahead:
Whether any of the the ten country bids lead to more signed deals and new nodes.
What happens when a node’s members and its host government differ in vision, values, and growth.
Succession and governance decentralization as network state projects scale
Whether Alatau City starts operations in 2027 as planned.
Ferguson’s bet on US talent inflows versus Balaji’s bet on the opening that tighter US visas create.
A year ago the question was whether network states could get onto land at all. They can, and the bottleneck has moved to people: leaders who run nodes in founder mode, and communities that build trust without the inherited axioms an organization like the RSS starts with. Network School now has capital, a host state, and a curriculum.
For anyone weighing a move, I’d treat nodes as prospective middle-term options rather than permanent destinations. Projects will keep pivoting and evolving, so the useful question is what you keep if yours does: your network, your credentials, your assets and your right to leave.
Last year I wrote that the network state was no longer a theory.
This year’s question is what a network becomes once a state signs on.
My answer for now: it stays a network as long as its members have collective mobility and individual rights to exit. It becomes a network state as long as it has truly secured land for the network.
