Momir Amidzic

Momir Amidzic@momir_amidzic · 13 天前观点输出

@figo_saleh always had an itch to build. Great to see him shipping. 🔥

103310

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Momir Amidzic@momir_amidzic · 13 天前观点输出

RT @jocyiosg: 听完这期 OFR，我发现自己对 @Morpho 和 @sparkfinance 都更 bullish 了。 过去聊 Crypto Lending，大多还是牛市加杠杆那套逻辑。现在情况有点不一样：真实的融资需求开始出现，监管逐渐放行，机构也终于能…

06015

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Momir Amidzic@momir_amidzic · 13 天前观点输出

A lot of nuanced insights from @shanandwater & @hexonaut here, I enjoyed recording it

21161.2K

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Momir Amidzic@momir_amidzic · 13 天前赛道影响

AI editor != AI writing The point of an article is to get opinion/insight out there. The fastest way to do so is to record your opinion and let AI edit and structure it into an article. I think this will become normalized as you are using computer to do calculation rather than doing it by hand. If you use writing as a process of forming an opinion then writing process itself can't be replaced, but if you have formed opinion just record it and let AI structure it. Using AI as an editor is a great advantage and I don't see a reason why resist it. In my opinion there is zero chance Stan let AI write article for him, he probably has formed opinion and can defend each paragraph of the writing. Also, the accuracy of Pangram detection is debatable.

103634

AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。

Momir Amidzic@momir_amidzic · 18 天前观点输出

Bessent is really a master of manipulation. Buying yen and pretending it’s yen defense when in fact it’s preventing Japanese to sell Treasuries into yen purchase and further stress on treasury market. Quite impressive tbh. But also shows how worried about Treasury market he really is.

000138

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Momir Amidzic@momir_amidzic · 2026/08/08观点输出

Agree I was thinking of writing Boring Crypto thesis as the next stage of consolidation but the core of the idea is presented nicely over here

103768

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Momir Amidzic@momir_amidzic · 2026/08/05观点输出

Oh wow This is a great one

102254

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Momir Amidzic@momir_amidzic · 2026/07/29市场影响

Today’s AI economics do not price in China’s coming impact on the cost of intelligence. Until recently, I assumed that America possessed a durable structural advantage in AI. It had the deepest capital markets, the leading research laboratories and the strongest position in advanced semiconductors. I no longer think that framework is sufficient. The AI race must instead be evaluated across six dimensions: chips, know-how, capital, energy, data and political will. America’s clearest remaining advantage is advanced semiconductors. Its advantage in know-how is much less obvious. I no longer believe that America enjoys a decisive advantage in the depth of its capital base either. America’s financial system is better at identifying and financing uncertain frontier innovation. China’s bank-led system is much better at flooding strategic industries with capital once the state decides that something must be built. That distinction matters enormously as AI shifts towards industrial deployment. China clearly leads in energy capacity and infrastructure. Data is roughly equal. China enjoys a clear advantage in political will and the ability to build at scale. So why has China not yet started a much larger AI infrastructure cycle? My view is that it may be waiting deliberately. Had China begun a massive build-out earlier, it would have been forced to rely more heavily on restricted NVIDIA hardware, thereby financing an American-aligned semiconductor and infrastructure ecosystem. By waiting, China may be giving Huawei, CXMT and the broader domestic semiconductor supply chain enough time to become viable. The trigger is therefore not semiconductor parity. It is a minimum viable domestic semiconductor system. Once that threshold is crossed, many of the other constraints begin to disappear. The capital is there. The energy is there. The engineering expertise is increasingly there. And the political capacity to co-ordinate a build-out is already well established. China has demonstrated this playbook before through property, high-speed rail, electric vehicles, batteries and manufacturing infrastructure. AI will be next. A Chinese build-out comparable in ambition to its previous industrial investment cycles would have enormous consequences for the supply and therefore the price of intelligence. The brutal risk for today’s AI economics is not simply that China produces a better model. It is that China turns intelligence into an industrial commodity and collapses its cost. https://t.co/vUTNR2UsE0

002146

AI：偏平台经营与生态扩张，强调交易平台和应用入口的长期位置。

Momir Amidzic@momir_amidzic · 2026/07/29观点输出

Generally speaking, the more distribution for the exchange, the better, e.g. builder codes. But there are several nuances, in my opinion: - Uniswap’s business model is not really aligned with this third-party distribution approach. Uniswap pool fees are likely going to trend towards zero, and Uniswap pool fee switch is limiting itself to the proportion of LP fees. You even have to defend taking only around 15% of pool fee for the protocol. The net take rate of Uni seems to be only around 1-2bps (net of LPs / total volume), which is extremely low for a spot exchange. Whether Uniswap has the leverage to change this is debatable. - For Uni token holders, therefore, the best hope is that Uniswap’s front-end owns the distribution because this is the flow that could be eventually monetized much more effectively at rates closer to the spot exchange market. I think these are key reasons why Uniswap’s reliance on third-party distribution is perceived as a weakness versus being celebrated as a win for many other protocols.

000142

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。

Momir Amidzic@momir_amidzic · 2026/07/20观点输出

1) Crypto is no longer the dopamine trade. The opportunity cost of allocating risk capital to crypto is very high right now. I don’t think this is structural, it’s temporary. The real question is how long it lasts. 2) Fewer data centers dedicated to BTC mining don’t really matter that much. I used to think they do, but no longer. Outside of extreme scenarios, Bitcoin’s hash rate is mostly meme metric in my opinion.

103425

AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。