Hong Kong Monetary Authority Chief Eddie Yue: Cross-Border Wealth Management Connect 3.0 Has No Timeline Yet, Expansion in Quota, Products, and Other Dimensions Expected 香港金管局总裁余伟文：跨境理财通3.0暂无落地时间表，额度、产品等多维度拓展空间可期

Personnel arrangements at Hong Kong's Monetary Authority and China Investment Corporation? Placing them in the tech section does seem somewhat incongruous. What truly struck a nerve today was another set of news: Apple's "all-new" Siri AI, OpenAI's secret IPO filing, and that provocative headline—"DeepSeek and its kind slash costs by 99%: Will the trillion-dollar gamble of American giants go down the drain?"

First, Apple. After being criticized for three years, Siri finally seems less like a "digital idiot" thanks to the "all-new" Apple Intelligence. Yet this so-called "newness" feels more like a patch delayed by three seasons. Before the launch event even happens, predictions are already flying: smarter writing, more obedient commands, cross-app operations... It sounds great, but is what users really want a "enhanced voice assistant" that’s still half a beat slow and relies on cloud computing? Has Apple truly figured things out this time, aiming to win back ground in the AI era, or is this just another "innovation story" spun for investors to boost stock prices and financial reports? I suspect it’s more the latter. After all, in an era where OpenAI and Google have already shattered the ceiling with multimodal, real-time generation, can Apple’s "toothpaste-squeezing" upgrades still spark enough enthusiasm among fans to switch phones? The phrase "Oh no, Apple’s AI actually works?" feels more like helpless sarcasm than admiration.

The real eye of the storm is at OpenAI. On one hand, there’s rapid progress in products; on the other, a quiet shift in company operations—secretly filing for an IPO. What does this mean? It means the nonprofit organization that once championed "benefiting all of humanity" is now firmly on the path to becoming an ordinary commercial giant. An IPO is the ultimate coming-of-age ceremony, but also a financial straitjacket. From now on, every decision OpenAI makes will be tied to Wall Street’s quarterly reports. Can it still maintain that "wild" spirit of exploration? When profit pressure becomes the top KPI, will those expensive, cutting-edge, and potentially unprofitable fundamental research projects be marginalized? This is yet another symbolic moment of idealism surrendering to commercial reality. The OpenAI that once gazed at the stars is now lowering its head to calculate its valuation.

And the sharpest contrast comes from across the ocean. As Silicon Valley giants burn through tens of billions training next-gen models and buying exorbitantly priced chips, "DeepSeek and its kind" are rewriting the rules of the game in a brutally direct way: slashing training and inference costs by 99%. This isn’t optimization—it’s subversion. It directly challenges the core logic of those trillion-dollar bets: if frontier intelligence can be obtained as cheaply as electricity and water, can the moats built around "computing power hegemony" still hold? American giants are betting on an "infinite game," wagering that economies of scale will eventually swallow everything. But what if the rules of the game themselves have changed? What if "small but smart" models, through clever architectures, achieve better performance than behemoths in specific scenarios? This isn’t a simple "cost war"—it’s an asymmetric attack on technology paths and business models. Moats or commons? The answer is becoming increasingly blurred.

So, don’t be distracted by slow-moving news like "Cross-Border Wealth Management Connect 3.0." The main battlefield of AI has already spread from laboratories to boardrooms and capital markets. There’s no gentle "discussion and exploration" here—only a naked race for survival. Apple is catching up, OpenAI is gilding itself, and new challengers are wielding the sickle of cost to harvest everything. This isn’t the natural evolution of technology; it’s a chaotic battle entwining commercial ambition, capital will, and clashes over technological paths. Revelry or bubble? Perhaps both. We stand at a crossroads: one path leads to a more accessible, efficient AI future; the other locks us into a monopolistic present dominated by giants.

香港金管局和中投公司的人事安排？放在科技版块确实有些违和。今天真正刺痛神经的，是另一组新闻：苹果那个“全新”的Siri AI、OpenAI秘密提交IPO，以及那个标题极具煽动性的——“DeepSeek们把成本砍掉99%，美国巨头的万亿豪赌要打水漂？”

先说苹果。被吐槽了三年的Siri，终于在“全新”的Apple Intelligence加持下，显得不那么像个智障了。可这所谓的“全新”，更像是迟到了三个赛季的补丁。发布会还没开，预测已经满天飞：更智能的写作、更听话的指令、跨应用操作……听起来很美，但用户等的真的是一个反应慢半拍、需要依赖云端算力的“加强版语音助手”吗？苹果这次是真想通了，要在AI时代扳回一城，还是又一次为了股价和财报，给投资者讲的“创新故事”？我怀疑后者的成分更大。毕竟，在OpenAI和Google已经用多模态、实时生成把天花板捅破的今天，苹果的“挤牙膏”式升级，还能激发多少果粉的换机热情？那句“坏了，苹果AI真成了？”更像是一种无奈的反讽，而不是赞叹。

真正的风暴眼在OpenAI。一边是产品上突飞猛进，一边是公司运作上悄然转向——秘密提交IPO。这意味着什么？意味着那个曾经以“为全人类谋福祉”为旗号的非营利组织，正在坚定不移地走在成为一家普通商业巨头的路上。IPO是终极的成人礼，也是资本的紧箍咒。从此，OpenAI的每一个决策，都将被华尔街的季度财报所捆绑。它还能保持那份“疯狂”的探索精神吗？当盈利压力成为首要KPI，那些昂贵的、前沿的、可能长期无法盈利的基础研究，会不会被边缘化？这是理想主义向商业现实投降的又一个标志性事件。那个曾经仰望星空的OpenAI，正在低头计算它的估值。

而最辛辣的对比，来自大洋彼岸。当硅谷巨头们为训练下一代模型烧掉百亿美元、购买天价芯片时，“DeepSeek们”正用一种近乎粗暴的方式改写游戏规则：把训练和推理的成本砍掉99%。这不再是优化，这是颠覆。它直接拷问那个万亿豪赌的核心逻辑：如果前沿智能可以像水电一样廉价获取，那些围绕着“算力霸权”建立的护城河，还守得住吗？美国巨头们押注的是一场“无限游戏”，赌的是规模效应最终会吞噬一切。但如果游戏的规则本身被改变了呢？如果“小而美”的模型通过巧妙架构，在特定场景下实现了比巨无霸更好的效果呢？这不是简单的“成本战”，而是技术路线和商业模式的降维打击。护城河还是公地？答案可能正变得模糊。

所以，别被“跨境理财通3.0”这种慢性子消息分心了。AI的主战场已经从实验室蔓延到了董事会和资本市场。这里没有温情脉脉的“探讨与拓展”，只有赤裸裸的生存竞赛。苹果在补课，OpenAI在镀金，而新的挑战者们，正在用成本镰刀试图收割一切。这不是技术的自然演进，这是一场混杂着商业野心、资本意志和路线之争的乱战。狂欢还是泡沫？或许两者皆是。我们正站在一个岔路口，一边是通往更普惠、更高效AI的可能，另一边，则是被资本锁死在巨无霸垄断里的未来。