NIO, XPeng, Li Auto Disclose May Delivery Data “蔚小理”披露5月交付数据

The nickname "NIO-XPeng-Li Auto" sounds like an unbreakable trio, but May's delivery data revealed it has already cracked into two-and-a-half. NIO delivered 37,705 vehicles in a single month, a staggering 62% year-on-year surge, as if someone had stepped on an electric accelerator. Meanwhile, XPeng saw a slight 4% dip, and Li Auto plunged 18.4%—the latter's decline is particularly stark, nearly falling a full car-length behind NIO. This isn't parallel progress at all; it’s the trio becoming "NIO leading the pack," with the other two more like passengers along for the ride. Mr. Li of Li Auto often talks about family, but data doesn’t care about sentiment—nor does the market. When consumers vote with their wallets, the once-glowing "dad car" halo seems to be fading. XPeng’s drop may not be large, but in this competitive atmosphere of comparing, learning, catching up, and surpassing, standing still is falling behind. NIO quietly pulled off a comeback—perhaps it strategically laid out battery-swap stations like a maze, luring all consumers inside.

While the new energy vehicle sector battles fiercely, AI has taken a surprising turn—into the pig pen. Muyuan and Alibaba Cloud have partnered to develop an "Intelligent Pig Farming Large Model." At first, it sounds like dark humor: humanity’s most advanced computing power and models are now being used to study what pigs eat, how to breed them, and whether they’re sick. But after the chuckle, this may be the most authentic form of AI implementation—cutting straight to efficiency, no fluff. A "Little牧 Assistant" has compressed health checks for 600 pigs from 20 minutes to mere seconds. This isn’t a percentage-point improvement—it’s a hundredfold leap. It’s far more concrete than many "revolutionary AI applications" that still exist only on PowerPoint slides. The pig farming industry is vast but technologically extensive, making it a ripe target for AI empowerment. Yet, the partnership announcement is crammed with trendy terms like "Qianwen Large Model" and "intelligent computing power," but fails to clarify which specific, headache-inducing pain point it actually solves. Can it precisely predict pork prices? Can it prevent African swine fever? Otherwise, this "strategic collaboration" risks becoming yet another meticulously orchestrated PR stunt. Pig farming still relies on experience and luck—AI just adds a high-tech backdrop.

Looking at these two events side by side is quite revealing. On one side, the electric vehicle market—a blood-red ocean—where giants compete fiercely over every digit in delivery figures. On the other, AI is seeking the most grounded, "earthy" industrial scenarios to prove its value. One is at the extreme frontier, the other deeply traditional, yet both ultimately point to the same core: technology must serve real commercial efficiency and cost calculations. NIO’s growth may stem from cumulative efforts in product iteration and service network, while Li Auto and XPeng’s declines could be the market’s ruthless questioning of product definition and iteration speed once the initial hype cools. As for AI in farming, unless it can truly reduce breeding costs and improve pork quality, it will struggle to have lasting vitality beyond capital stories. The market doesn’t care how advanced your model is—it only cares whether your pigs are healthier and more profitable.

So, don’t be dazzled by those flashy tech buzzwords. Whether building cars or raising pigs, the ultimate comparison is about who can twist technology into tangible productivity and apply it directly to the user’s (or farmer’s) pain points. Otherwise, sliding data and partnership announcements are nothing more than fleeting ripples in an industry bubble.

“蔚小理”这名字，听着像是个铁三角，但5月的交付数据一出来，这铁三角分明已经裂成了两角半。蔚来一个月交付37705辆，同比暴涨62%，这增速简直像是踩了电门。而小鹏和理想呢？一个小跌4%，一个大跌18.4%，理想的跌幅尤其刺眼，几乎快被蔚来甩出一个身位。这哪是齐头并进，分明是“蔚小理”变成了“蔚一家”，后面两位更像是在陪跑。理想的李总总爱谈家庭，可数据不讲情面，市场更不讲——当消费者用钱包投票时，曾经的“奶爸车”光环似乎有点褪色了。小鹏的跌幅虽然不大，但在这三家比学赶帮超的氛围里，原地踏步就是退步。蔚来倒是闷声打了个翻身仗，不知道是不是把换电站铺成了迷魂阵，把消费者全绕进去了。

这边新能源汽车在卷生卷死，那边AI却一头扎进了猪圈。牧原和阿里云合作搞“智能养猪大模型”，乍听有点黑色幽默：人类最前沿的算力和模型，最终去研究猪吃什么、怎么配种、有没有生病。但笑完细想，这才是AI落地最真实的模样——不玩虚的，直接抠效率。一个“小牧助手”把600头猪的健康检测从20分钟压到秒级，这效率提升不是百分比，是百倍。这比很多还在PPT上飞舞的“颠覆性AI应用”实在多了。养猪产业规模巨大，但技术粗放，确实是AI赋能的好靶子。不过，合作新闻里堆满了“千问大模型”、“智算算力”这些时髦词汇，却没提清楚究竟解决了哪个最具体、最头疼的痛点。是能精确预测猪价，还是能防住非洲猪瘟？否则，这“战略合作”很容易沦为又一场精心策划的PR活动，养猪还是得靠经验和运气，AI只是多了个高科技背景板。

这两件事并置来看，挺有意思。一边是已经杀成红海的电动汽车市场，巨头们在交付数字的毫厘间拼杀；另一边是AI在寻找最朴素、最“土味”的产业场景去证明价值。一个极度前沿，一个极度传统，但本质上都指向同一个核心：技术最终要服务于真实的商业效率和成本计算。蔚来的增长可能来自产品迭代和服务网络的厚积薄发，而理想和小鹏的下滑，或许是市场热度消退后，对产品定义和迭代速度的残酷拷问。至于养猪AI，除非它能真正降低养殖成本、提高猪肉品质，否则在资本故事之外，很难有持久的生命力。市场不关心你用了多先进的模型，只关心你的猪是不是更健康、更赚钱。

所以，别被那些光鲜的科技名词唬住。无论是造车还是养猪，最后比的都是谁能把技术拧成一股实实在在的生产力，拧到用户（或者养殖户）的痛点上。否则，数据下滑和合作发布会，都不过是行业泡沫里短暂的水花。