Investment Banking → Hedge Fund → Entrepreneur Love Research, Love Sharing, Love Family

The global refining market is experiencing persistent supply tightness, with refined product rallies significantly outpacing crude oil—particularly in the diesel sector. Goldman Sachs has more than doubled its 2027 diesel crack spread forecasts, raising the US estimate from $27 to $63 per barrel and the EU estimate from $19 to $49 per barrel. The refining margin (or crack spread) represents the price differential between raw crude oil and the refined products it yields, such as diesel and gasoline. It serves as the primary benchmark for a refinery's processing profitability. Several key factors are driving this upward momentum: 1. Structural Supply Contraction in Europe Over the past few years, Europe has continuously shed refining capacity. Driven by the European Green Deal, soaring carbon emission allowance costs, and the need to hedge against the long-term demand destruction anticipated from electric vehicle (EV) adoption, a wave of economically marginal refineries has been forced to close or pivot operations. Crucially, underlying demand is not declining as rapidly as capacity. This structural deficit is currently exacerbated by global unplanned refinery outages running 60% above normal historical levels. 2. Geopolitical Conflicts Geopolitical tensions have further compounded the issue, with strikes on Middle Eastern and Russian refineries disrupting both infrastructure and logistics. As a result, geopolitical risk premiums are spilling over from upstream crude into the downstream products market. Additionally, Russia's extended ban on diesel exports has further restricted global supply. 3. Seasonal Demand Surges Coupled with the upcoming peak agricultural season in Brazil and rising seasonal heating demand in the Northern Hemisphere, the supply-demand imbalance is set to intensify. With European diesel inventories already structurally low, any minor supply-side disruption is heavily amplified. Reuters previously noted that European diesel refining margins briefly spiked to historical highs of approximately $65 per barrel. If diesel prices remain elevated without a corresponding surge in crude, widening diesel crack spreads will pad refinery margins, theoretically acting as a major tailwind for refinery equities. However, investors must remain vigilant regarding windfall tax risks. Historical precedent—notably during the 2022 energy crisis—demonstrates that surging downstream profits often prompt European and other regional governments to levy excess profit taxes on energy firms, introducing substantial policy risk for related investments. To mitigate rising energy procurement costs, market participants typically deploy the following institutional hedging instruments: Strategy 1: Long Gasoil/Diesel Futures. Taking a long position in gasoil or diesel futures provides a direct, linear price hedge. As physical procurement costs rise due to surging diesel prices, the mark-to-market profits from the futures position offset these operational expenses. Strategy 2: Purchasing Diesel Call Options. For those primarily concerned with extreme, sudden price spikes (tail risk), call options offer a superior alternative to outright futures. By paying an upfront premium, buyers cap their exposure to extreme upside risk, effectively locking in a maximum purchase price ceiling. If diesel prices fall instead, the downside loss is strictly limited to the premium paid, completely avoiding the cash-flow drain of daily margin calls associated with futures contracts. Looking Ahead: A New Normal? This dynamic raises a critical, long-term question: Will the pace of European refinery capacity contraction consistently outpace the decline in underlying diesel demand? If the answer is affirmative, the elevated refining margins we are currently witnessing are not merely the result of a transient energy shock. Rather, they signify a "new normal"—a permanent structural shift resulting from Europe's evolving energy architecture.
Diesel Crack Spread soars to highest level in history 📈 📈
1
2
677
Captain Teemo on duty 🫡🍄 Time to scout the bush. @alloco_fi 🤔🤔
Many high-growth companies stay private for longer. Access to them still tends to depend on the right fund, channel or relationship. @alloco_fi structures investment exposure as on-chain products: > ETF Tokens are live today; selected pre-IPO exposure is planned next > Issuance, rebalancing and distribution handled on-chain The team reports 12 live products and more than $40M in AUM. Catch @Captainteemo3 at EASY Residency S4 Demo Day.
1
2
5
680
这几天跟国内几家半导体产业方,以及几个做私募的朋友聊下来,一个越来越强烈的感受是:中美这两个全球化体系中最重要的供给方与需求方,正在殊途同归。逆全球化趋势不可逆转,曾经建立在全球资本自由流动基础上的科技投资框架正在被国家安全、产业政策和地缘竞争重新定义。 Maybe PLTR was directionally right after all. 国内半导体产业的资本路径现在很清晰。资金来源本土化,投资主体产业化,资本期限长期化,政府资金向更高层级集中。 国办54号文件和国资委46号令一套组合拳下来,政府投资基金开始收缩战线。县级基金整合,资源往省级母基金和大型产业基金上集中。到了核心科技资产的一级市场的战略配售层面,过去市场化PE/VC、美元基金都能广泛参与的格局也在改变。 现在真正有资格坐上牌桌的,主要是产业资本、省级产投和大型国资平台。稀缺的明星项目,还会引入社保、险资这些长期资本。 我自己的理解很直接:化债。肉来了,先给穷亲戚们分一分。 地方过去最重要的资产化机器是土地。现在家印同志们在铁窗泪,房地产这台机器基本熄火;另一边,社保和养老金的长期压力,看生育率也大概知道。 明星科技公司的上市,开始承担比融资本身更多的功能。即重新创造一批能够承接公共部门长期资本的优质权益资产。 相应地,纯财务投资人的位置越来越尴尬。 强监管下,IPO前新增股东、异常定价、多层嵌套SPV、资金来源以及最终权益持有人,核查越来越细。过去FA那套Pre-IPO老股转让、撮合退出的生意越发举步维艰。此处给曾经风光无两的华兴资本掬一把泪。 更麻烦的是持有期限。 现在核心项目优先考虑长持资金。 DeepSeek最近一轮融资就是一个例子。除国家人工智能产业投资基金外,外部投资人进入梁文锋控制的有限合伙,没有投票权,锁五年。而我自己给客户拿到的Pre-IPO 基金额度更夸张,要求LP锁十年。 至于境外资本很难再获得有效参与机会,简单说来就是在清退美元基金。 标志性事件,自然是Meta收购Manus。 今年3月,Manus CEO肖弘和首席科学家季逸超被要求回到北京,就Meta收购交易接受监管问询,暂时不得离境。 几乎同一时间,路透社报道红筹架构赴港上市的监管口径收紧,企业要先拆红筹,再递表。现下,月之暗面就在紧锣密鼓地拆境外架构。 这让很多科技企业傻了眼。当初为了拿美元资金搭红筹,现在想回来上市,又得把红筹拆掉,一根筋成了两头堵。 从3月至今,只有初速度科技Momenta成了特例。历史股东里美元基金和境外资本占比很高,客户又主要是海外车企,最后却保留开曼架构成功登陆港股。大概是因为这是物理AI这条赛道结出来的一颗硕果。 先A后H也成了趋势。科技企业要先在科创板完成彻底的股东穿透、资本属性识别和股权结构锁定,再去港股融资。 这些变化叠加以后,国内核心科技资产的定价方式发生变化。 过去一级市场有大量财务投资人,他们有明确的ROI、基金期限和退出要求,会围绕估值、上市时间和二级市场流动性计算IRR。项目贵不贵,自然有人用钱投票。财务资本提供的不只是资金,也提供价格发现和流动性。 但进入核心项目的钱越来越集中在产业资本、地方产投和长期资金手里,一级市场的边际买家减少,资金属性也更趋同。 到了二级市场,这套逻辑还在继续。 长鑫为例。上市首日,长鑫无限售流通股只占总股本约6.7%,大量战略配售股份锁定12到36个月。一个市值3万亿元级别的公司,最初的价格,是在极小的自由流通盘上形成的。 按朋友那边了解到的情况,长鑫上市前后,监管一个重要考虑就是避免这种mega IPO在短期资金推动下迅速冲到过高估值,对整个科技板块造成巨大的资金虹吸和波动外溢,重蹈中芯国际当年覆辙。 监管预期先让市值在3万亿元附近完成初步定价,再逐步交给市场。上市前几天,部分公募基金经理在二级市场的主动买入受到限制。随后韩国市场去杠杆,A股科技板块跟着大幅调整,等筹码和交易结构稳定下来,机构买入逐步放开。到8月17日,市值突破4.1万亿元。 到这里,可以明确看出监管职能扩大化。从IPO审核,进一步延伸到市场微观结构。对核心战略资产的融资到定价,监管已经形成一套完整的资本结构和流动性管理。 市场这只无形的手发挥的作用变小,负责金融稳定、产业安全和资本秩序的有形大手作用在变大。 宇树之后,监管又开始收紧亏损科技企业的上市空间。 细分行业前三还有机会,第三第四差距不大,那第三名自己也得歇菜。当然,财联社说他问了,券商们说没这回事,没听说。 至于是不是事实,看下具身智能企业未来上市情况遍知一二。 回头再看Alex Karp这些年反复讲的那套东西,竟在一点点成为现实。 全球化退潮之后,国家重新成为科技产业最重要的组织者和买方。AI、半导体、能源、先进制造被纳入国家安全的框架;资本也开始有国界,有身份,有期限,有任务。
1
17
36,204
Season 4 done, onwards. 🎓 这次最幸运的事,就是遇到Ella姐@ellazhang516。必须最大声感谢,并致以敬意🫡🫡! 几轮交流下来,Ella姐一直从专业投资人的视角,帮我们重新梳理商业叙事、产品价值,以及一个项目真正的核心到底是什么。 我最大的收获,是开始慢慢学会跳出纯执行的视角。不再只是想着“这个东西怎么做出来”,而是开始试着从投资人的角度重新看一个项目:为什么是这个市场,为什么是现在,为什么是我们。 最后几天,大佬投资人们也纷纷降临。和他们聊完之后,我也更能体会CZ @cz_binance 坚持做YZiLabs Residency的良苦用心。 早期项目本来就充满不确定性,最后真正能跑出来的也一定只是少数。正因为这样,愿意在这么早的时候给创业者资源、机会和信任,本身就是一件很难得的事。 想想还挺感动的。那就继续努力,别辜负这份信任!💪 感谢一路上遇到的每一个人,也感恩这段不丹之旅🏔️ Building starts here! 🚀🫡 敬请期待Alloco @alloco_fi ,Coming Soon 👀🔥
EASY Residency was never easy. The building, the pivots, the long days — none of it was. But believing in how far these founders can go? “That was EASY.” 😁 Congrats, Season 4 graduates 🎓
4
1,280
这次关于 OpenRouter 的讨论让我印象深刻👏 @tomhschmidt @tarunchitra 随着模型推理逐渐商品化,价值捕获可能会进一步向Routing和Order Flow迁移。 而当不同服务提供商开始围绕路由机制进行策略博弈时,我们会发现,很多传统金融市场和Defi里非常熟悉的问题也随之出现:MEV、逆向选择(Adverse Selection)、激励机制设计,以及更复杂的市场博弈。 走到这一步,它已经不再只是一个 AI 基础设施问题,也并非单纯的 Token 机制问题。 本质上,这是一个非常传统的华尔街市场结构问题,只不过正在以 AI 和链上市场的新形式重新出现。
Crypto’s finally cool again. Meanwhile, AI could learn from DeFi’s past. > Haseeb claims AI just removed bureaucratic speed bumps. > Robert sees a path for small token raises. > Tom thinks the SEC is addressing 2017 problems in 2026. > Tarun believes AI inference markets could be crypto's next cycle. Timestamps 00:00 Intro 01:22 Tom & Tarun meet the King of Bhutan 04:13 Bitcoin rebounds and crypto gets hot again 08:10 What a compliant U.S. Hyperliquid could look like 13:14 The SEC's proposed Regulation Crypto Assets 23:53 Are the new token rules eight years too late? 28:23 AI removes the government's bureaucratic speed bumps 31:12 Why AI inference tokens could power the next cycle 34:13 Stripe buys OpenRouter and AI starts looking like DeFi 43:28 Router economics, data markets, and new security risks 51:01 Bull-market vibes from Bhutan and Asia 🔥Stay updated with all the latest hot takes by following and subscribing to @_ChoppingBlock and @unchained_pod! 🎥 YouTube: piped.video/M2oUa-l9gBY 🎧 Spotify: bit.ly/3wiIOyy 🍎 Apple: bit.ly/3w9HQ7J 🎙 Podcast Home: choppingblock.xyz
7
3
853
Thanks Coco!谢谢大家一路陪伴!@yzilabs @ellazhang516 @cz_binance 谢谢鼓励和指导!继续成长! @alloco_fi
3
658
Very excited to host the event! It was a wonderful night!
Today, Alloco welcomed valued friends from across the crypto industry to a private dinner at @Amanresorts in Bhutan 🇧🇹 This is Alloco’s first event in Bhutan, bringing together familiar faces and new ones. We shared a beautiful evening together, filled with great conversations and memorable moments.
1
4
343
Bravo🔥🔥🔥!
Today, Alloco welcomed valued friends from across the crypto industry to a private dinner at @Amanresorts in Bhutan 🇧🇹 This is Alloco’s first event in Bhutan, bringing together familiar faces and new ones. We shared a beautiful evening together, filled with great conversations and memorable moments.
9
2,738
Grateful to have @alloco_fi be part of EASY Residency Season 4. It’s been an incredible journey building alongside such a strong group of founders. Thank you @yzilabs for the support and belief in what we’re building. Just getting started. 🚀
2
13
1,863
👀
In two days, @EASYResidency S4 founders take the stage for Demo Day. 🇧🇹 This summer, we came to Bhutan — in the heart of the Himalayas — to build across stablecoins, payments, DeFi, AI, and beyond. Here’s a glimpse into the founders’ journey with YZi Labs.
7
750
懂了👍。哈哈哈哈有人说百度从来都是起个大早赶个晚集,目光如鹰,行动如树懒。
gemini会按照搜索引擎来,所以你还是得往搜索引擎里砸钱,换句话说你要去弄tag去专门做gbp在网页上 怎么说呢,方便不少 百度他结合得很差……
1
2
791
Neocloud想扩规模都会发债。之前笑CRWV BS丑陋的回旋镖来了。 云计算本身就是重资产行业,资产负债表再丑陋,IDC都得建。只不过AIDC对资本和算力的吞吐大到抽象,让市场第一次意识到,过去被看作“现金牛”云厂商,Capex可以膨胀到什么地步。 除了Capex,客户集中度也是新云被诟病的一点。CRWV被杀估值的理由就是高负债和大客户集中度。但现在基本都在往这个方向走。 原因很简单,nbis以前的客户偏SMB,分散度虽高,但短单不稳定且价格敏感。对于处于高速扩张期的新云来说,这种客户结构未必比大客户更优。 其次,scale到大量IG级客户,拿到长期合同,合同可以成为融资资产(凭借IG级客户信用)。通过SPV隔离母公司风险,再以长期合同现金流和底层资产举债,融资效率和资本结构都会发生变化。 不管是走AI Factory $CRWV 还是Full stack AI cloud的 $NBIS ,最后都会走到excution to scale上。
Nebius $NBIS just announced plans to raise $4.5 Billion by selling convertible senior notes
1
406
这就是我上一篇文提到的逻辑。TrendForce预计27年,DRAM的供给缺口还会进一步扩大。以HBM为例,限制供给释放的不只是前道产能,还有先进封装能力和良率爬坡。从宣布扩产、建厂,到高良率量产,中间存在明显的时间滞后。 但市场目前并没有充分给这部分基本面定价,近期交易更多还是在消化前期去杠杆冲击,并反复博弈存储价格与盈利周期是否见顶。这反映在估值上,海力士的fwd PE明显低于美光这个海外同行。 于是海力士选择股票回购来提振股价。P=PExEPS 当PE倍数难以上行时,股价的主要驱动力就从PE扩张转向EPS增长;在经营利润增长之外,股票回购通过减少在外股份,可以进一步放大每股盈利的增长。
最近发现一个规律,只要每次 SK 海力士跌 10%,公司就会发一篇利好消息。 从大老板崔泰源,7 月底自掏腰包买股票,再到 100 万亿股东回报计划,今天尾盘连回购燃烧股票的消息都出来了。。
2
352