Minulý týden mi kamarád řekl o AI produktu, který vyvíjí.
Chtěl ověřit všechno on-chain. Každou inferenci, každý výstup, s co největším možným důkazem.
Zepšel jsem se: "Je něco na tom seznamu, co jsi se rozhodl neověřovat tímto způsobem?"
On zmlknul.
To může být jedna z těžších otázek v AI x crypto.
Většina AI-blockchain projektů nepadá, protože je kryptografie špatná. Padá, protože ověřují všechno stejným způsobem, dokud nic neběží dost rychle na to, aby se to dalo skutečně používat — a v té době si toho nikdo nevšimne.
Proto považuji architekturu @OpenGradient HACA za zajímavou.
Důkazy s nulovými znalostmi jsou jeden příklad. Důkaz ZKML může být 1 000 až 10 000krát pomalejší než spuštění modelu — vlastnost kryptografie, ne něco, co by mohl OpenGradient optimalizovat.
Místo toho se zaměřují na to, co mohou kontrolovat: specializaci uzlů HACA, spektrum ověřování TEE/ZKML, bránu x402, MemSync a Model Hub.
Na první pohled to vypadá jako kompromis. Ale je to těžší disciplína: vědět, které části skutečně potřebují být bez důvěry, místo aby se automaticky spoléhali na to, co zní nejpůsobivěji.
Zdržení nezaručuje přijetí. Ale většina neúspěchů AI-crypto nevznikla z slabé kryptografie — vznikla z toho, že udělali všechno maximálně bez důvěry, dokud to nebylo příliš pomalé na to, aby se na tom dalo stavět.
To samé platí pro investování. Jsme přitahováni k čemukoliv, co zní technicky maximálně. Ale větší riziko je podpořit tým, který ještě nenašel tu hranici.
Možná to je to, co OpenGradient opravdu testuje s HACA. Nezda, zda mohou ověřit více — ale zda přesně vědí, co to potřebuje.
Infrastruktura je kapacita. OpenGradient neprodává kapacitu. Prodává ověřitelnost.
Každé vyhodnocení produkuje kryptografický důkaz. Model běžel. Výsledek je správný. Uloženo na řetězci.
To má význam v určitých oblastech — Chytré smlouvy reagující na výstupy AI. Autonomní agenti, kteří potřebují auditovatelné rozhodnutí. Protokoly, které nemohou důvěřovat centralizovanému API.
To je menší trh než „všechno AI výpočetní“. Je to také trh, který nikdo jiný nevytvořil.
2M vyhodnocení před TGE. 500K ověřených důkazů. 2 000 modelů naživu. Aplikace už jsou v produkci. $9,5M od a16z, Coinbase Ventures. 12měsíční klauzule, než mohou insiders manipulovat s nabídkou.
$OPG spuštěno za $0.48 v dubnu. Minimálně se dotklo dna minulý týden.
Dříve jsem si myslel, že sázka na AI infra byla o růstu výpočetního výkonu.
Teď si myslím, že sázka je užší a konkrétnější: Stane se ověřitelná on-chain AI nutností, nikoli vlastností?
Pokud ano — OpenGradient je na začátku nevyužité kategorie. Pokud ne — je to dobře postavený produkt pro malý trh.
I used to think multi-asset restaking was mostly a distribution play for protocols like $BR .
More supported assets = wider audience. Simple.
That assumption feels incomplete now.
I’ve watched multiple restaking protocols launch with broad asset support early… but eventually the same problem appeared. Capital flowed in during incentive periods, then quietly rotated out the moment yields compressed elsewhere. The asset list grew. The sticky capital didn’t.
No real cross-asset utility. No compounding reason to stay. No economy forming underneath the yield.
So now I look at something else.
Interconnection.
Not the technical kind — the economic kind.
Does supporting multiple assets actually create relationships between them inside the protocol? Does BTC restaker behavior affect ETH restaker outcomes in meaningful ways? Can the system build interdependency between assets, not just host them side by side?
Because without interconnection, multi-asset support is just a feature list.
And without an economy forming underneath, restaking stays a yield product instead of becoming infrastructure.
That’s the layer I’m starting to watch more closely with $BR .
Not enough to call it solved. But enough to stay interested.
Still approaching it carefully.
Just watching whether the assets inside start to interact… not just coexist.
But after watching a few cycles, I’ve started paying less attention to what a network can do and more attention to what people choose to build on it.
That distinction matters.
Because technology advantages fade faster than most expect.
What tends to last is trust from builders.
The reason Bedrock keeps showing up on my radar is that the thesis feels less about winning attention and more about becoming dependable infrastructure.
Not the most exciting story.
But infrastructure rarely wins by being exciting.
It wins when people stop questioning whether it will be there tomorrow.
I’m still treating $BR as a trade.
Just starting to think the real signal isn’t the chain itself — it’s whether builders keep choosing it when nobody is watching.
OpenLedger and the Problem of Building for a Future That Hasn’t Arrived Yet
One thing I’ve learned from crypto is that being early and being wrong often look identical for a very long time. That’s what makes $OPEN difficult for me to think about. Because OpenLedger feels like it’s building around a future that makes sense in theory, but isn’t fully visible in practice yet. And that’s an uncomfortable place to be. Most markets reward solving today’s problems. OpenLedger seems focused on tomorrow’s problems. Ownership of AI outputs. Coordination of contributors. Value distribution across intelligence networks. These conversations feel increasingly important. But are they important enough today? I’m not sure. That’s the tension. The more I use AI, the more I understand the long-term argument. Intelligence is becoming infrastructure. People are integrating AI into work, research, writing, software development, and decision-making at a remarkable pace. Something fundamental is changing. But when I look at actual user behavior, I see something else. Most people aren’t thinking about ownership. They’re thinking about utility. They don’t ask who owns the model. They ask whether the model works. And that’s a very different incentive structure. It creates a strange challenge for projects like OpenLedger. The thesis may be correct. The timing may not be. Or maybe the timing is exactly right and the market simply hasn’t recognized it yet. That’s the part nobody can know. I keep noticing how many decentralized AI discussions assume awareness naturally follows importance. But history doesn’t really support that. People can depend on systems for years before questioning who controls them. Cloud infrastructure. Search engines. Social networks. The ownership conversation usually comes later. Much later. Often after dependency has already formed. That possibility keeps pulling me back toward $OPEN . Because if OpenLedger is right, it’s effectively trying to build the coordination layer before the ownership debate becomes unavoidable. That’s ambitious. And risky. Infrastructure designed for future demand always carries that risk. You can arrive too early. You can build before the market is ready. You can solve a problem people haven’t felt strongly enough yet. Still, there’s another side to this. If you wait until the problem becomes obvious, the opportunity may already belong to someone else. That’s what makes infrastructure investing so uncomfortable. The signals are rarely clear. You end up evaluating possibilities more than realities. And OpenLedger feels like one of those projects. I don’t look at $OPEN and see certainty. I see a question. What happens if AI becomes deeply embedded in economic activity, but ownership and value capture remain concentrated in a handful of places? Maybe that becomes one of the defining issues of the next decade. Maybe users never care enough for it to matter. Right now, both outcomes feel plausible. And that’s why OpenLedger still feels unfinished to me. Not as a project. As a thesis. The future it’s building toward hasn’t fully arrived yet. Which makes it incredibly difficult to measure — and impossible to dismiss entirely. #OpenLedger @OpenLedger $OPEN