Cursor的明天，是奇迹还是清零？

#

1.Cursor的未来
Cursor的命运，是被某个大厂吞并，融入他们的生态，还是能独立打出一片天地？说实话，我对Cursor的未来还是偏悲观。首先，做模型的人才远比做工具产品的人稀缺得多，其次，成本实在难以承受。每一次大模型的升级，调用费用都在直线上升。模型不给力，直接导致返回的代码质量更差，而创业公司的tokens消耗却居高不下！我自己每天用Vibe Coding处理些文字工作，感觉成本就已经超过了所有其他软硬件的总和。 代码所代表的市场容量依然巨大，而且很容易迁移到一些与代码近似的领域，比如数据分析、行业研究、金融等。

2.犹太人掌控金融，华人主攻AGI
这句话是听张小珺最新一期时听来的，但身边朋友的创业案例和我最近观察到的许多现象，确实越来越符合这个总结。一方面华人数量众多，另一方面华人足够努力。AI时代的公司并不完全靠堆人力，虽然有些场景确实需要一帧一帧地打磨，但那属于已有场景。 在全球化步伐如此迅猛的今天，未来一定会有更多新机会、新工具和新场景涌现。

3.ChatGPT 的优势
10亿DAU和模型极限，这是两条完全不同的赛道，而能在这两条赛道同时跑在前列的团队，屈指可数。这些优势包括品牌信任、日常习惯养成、产品设计、生态网络效应、全球运营能力、跨文化内容适配等。甚至一次简单的版本更新，可能就足以打碎许多创业公司的前景。 于是Anthropic更专注于代码这条赛道，谷歌也不甘示弱，迅速推出了Gemini-2.5-pro。也许用不了这个月底，又会有新的大招放出——毕竟GPT-5发布都让人觉得仿佛已经过去很久了。

4.大家都在关注产品
这是个相当务实的思路。毕竟做大模型已经不是一般公司能投入的了，即使能投入，又能否养大呢？ 大家一边选择产品，一边也不免感叹：“有了淘宝，谁还会做电商？”可谁又能想到后来拼多多的崛起，Shein的海外电商也是一夜之间冒了出来。无论是VC还是创业公司，抑或从业者，都在关注：什么样的公司和形态会在即将到来的AGI时代胜出，谁又能大鹏一日同风起，扶摇直上十万里？

大白话说，大家都在找机会，但能赚钱的巨头都在做，不赚钱的要么像Cursor一样成本高企，要么就是盲人摸象。

5.投资机会
许多人看好谷歌，认为谷歌的平台、广告、硬件，乃至一切，在AGI时代都极具吸引力，觉得AI的当下一定能再造辉煌。 我也有类似的看法。谷歌基本亦步亦趋，虽然代码水平没有超过Claude Code，智能也赶不上GPT-5，但不可否认，每次其他模型访问出问题时，都是Gemini来救火。不管是AIStudio、Notebook，还是后端Vertex接口服务，都是速度最快、智能尚可的选择。

不过，其实我还看好另一家公司，这正呼应了投资大佬阿段说的“敢为天下后”的公司——苹果。 Apple doesn’t rush to be first（苹果不抢第一个吃螃蟹的人）。前有iPhone发布前的Palm、BlackBerry、Nokia，后有AirPods发布前层出不穷的蓝牙耳机。历史一次次证明，苹果凭借这样的节奏，既能避开先行者的阵痛，又能在用户心中保持“进场=引领潮流”的光环。无数次技术萌芽期发展到苹果入场期，苹果都能卡位成功。这次AI发展看似掉队，实际上苹果依然保持着那熟悉的节奏。

当然，这一切都只是猜测。再等十几天，苹果新机发布会，或许就能初见端倪了。

6.苹果的新AI策略，会不会成为致命一击？

整理了下历年苹果发布会的时间和软硬件产品，虽然不知道下一代 AI 更成型的产品会是啥，但比如： 更智能的 AirPods 更强大的智能 iPhone 再收购下 Thinking Machines Lab 类似的 AI 创业公司？ 更实惠更注重隐私的多个大模型整合，等等

不管他人如何看，但至少笔者肯定会买单的。

Cursor, AGI, and Apple’s Calculated Strike: Navigating the Next AI Wave

#

1.Cursor at the crossroads
Will Cursor be folded into some tech behemoth’s ecosystem, or can it carve out a life of its own? I’m leaning pessimistic. Engineers who can build frontier models are far rarer than those who can ship tooling, and the math is brutal: every time an LLM levels up, the call price spikes. When the model is weak, the code it returns gets worse—yet a startup’s token burn keeps climbing. My own bill for running Vibe Coding on mundane text each day now eclipses the cost of every other piece of software and hardware I use. Still, the coding market is vast, and it can spill naturally into look-alike arenas—data analysis, industry research, finance, and so on.

2.“Jews run finance, Chinese chase AGI”
I caught that line on Zhang Xiaojun’s latest podcast, and the pattern shows up everywhere I look: a flood of Chinese founders, working absurdly hard. In the AI era you don’t simply pile on manpower; yes, some tasks still demand frame-by-frame polish, but those belong to yesterday’s use cases. With globalization racing ahead, brand-new opportunities, tools, and scenarios will keep bubbling up.

3.ChatGPT’s double moat
A billion DAU and the bleeding edge of model capability are two completely different tracks—and almost no team leads both. The real moat now includes brand trust, daily-habit lock-in, product craft, network effects, global ops, and cultural localization. A single version update can wipe out a fleet of startups overnight. Anthropic is doubling down on the code lane; Google, unwilling to trail, fired back with Gemini-2.5 Pro. Give it to the end of the month and another shockwave may land—GPT-5 already feels like ancient history.

4.Product, product, product It’s the only pragmatic stance. Building a frontier model is beyond most firms, and even if you can afford it, can you really nurse it to scale? People once scoffed, “With Taobao in place, who’d start an e-commerce site?” Yet Pinduoduo rose, and Shein’s cross-border store seemed to appear overnight. VCs, founders, rank-and-file alike are asking which shape of company will win the AGI era—who will, like the mythic roc, catch a single gust and soar for ninety thousand li. Bluntly put: everyone is hunting for daylight, but the profit zones are already occupied by giants. The rest either face Cursor-sized cost mountains or grope around in the dark.

5.Where to place a bet
Plenty of people favor Google. They see the platform, ads, hardware—the whole stack—as poised for another golden run in the AGI age. I’m sympathetic. Google moves in lock-step: its code lags Claude Code, its intelligence trails GPT-5, yet whenever rival APIs sputter, Gemini is the fire brigade. From AI Studio and Notebook to the Vertex backend, Google stays the quickest, smart-enough option. But I’m also eyeing another player—what investor Mr. Duan calls the company that “dares to be last”: Apple. Apple never rushes the first bite of the crab. Before iPhone there were Palm, BlackBerry, Nokia; before AirPods, a sea of Bluetooth buds. By entering late, Apple avoids pioneers’ growing pains yet still owns the spotlight. Time after time, the tech sapling grows tall, Apple strolls in, and somehow it’s suddenly Apple’s forest. AI looks no different: they seem behind, but the cadence feels familiar.

Ten days from now the new-iPhone keynote may offer our first real hints.

6.Could Apple’s AI play be the knockout blow?

Year  iPhone launch  Two headline acts (beyond the phone) 2024 iPhone 16 1. Apple Intelligence 2. Apple Watch X (all-new design) 2023 iPhone 15 1. Watch double-tap gesture 2. Full switch to USB-C 2022 iPhone 14 1. Dynamic Island 2. Apple Watch Ultra 2021 iPhone 13 1. Cinematic Mode 2. iPad mini 6 (edge-to-edge) 2020 iPhone 12 1. MagSafe ecosystem 2. HomePod mini

Glancing over Apple’s timeline, the next AI-forward gadget could be anything:
• Smarter AirPods
• A truly intelligent iPhone
• Snapping up a lab like Thinking Machines
• Cheaper, privacy-centric bundles of multiple LLMs

Whatever shows up, I’m ready to swipe my card.