10月22日凌晨，特斯拉的第三季度财报摆上了台面。总营收280.46亿美元，和去年同期的280.95亿几乎没动；净利润7.87亿美元，同比掉下去43%；毛利率19.6%，比去年的18%略好，但营业利润已经腰斩到7.2亿美元。成绩单不算难看，可汽车这门生意明显不长了。
In the early hours of October 22, Tesla's Q3 report landed. Total revenue was $28.046 billion, barely moving from $28.095 billion a year earlier; net income $787 million, down 43% year on year; gross margin 19.6%, a touch above last year's 18%, but operating profit had halved to $720 million. The print was not ugly, yet the car business had clearly stopped growing.
卖车这一块在拖后腿。三季度汽车收入205.6亿美元，同比下降3%，均价下滑和交付减少是主因。本季交付48.65万辆，虽然超出市场预期约5%，但同比还是少了2.1%；生产46.4万辆，同比增长4%。未交付订单堆到近四年最高，可电池供不上，工厂造不出足够车去填那些积压的单子。
The car side dragged. Auto revenue was $20.56 billion, down 3% year on year, hit by lower average prices and fewer deliveries. Deliveries were 486,500, about 5% above expectations but still down 2.1% annually; production 464,000, up 4%. Backlog hit a four-year high, yet battery supply fell short and factories could not build enough to fill those stacked orders.
真正的好消息在现金流。本季自由现金流15.3亿美元转正，市场预期原本是负49.9亿。资本支出47.3亿美元，同比猛增110%，却远低于预期的79.5亿，全年开支指引还从250亿砍回200亿。马斯克把战场从红海卖车，挪到了还握着绝对优势的AI和太空。
The real good news was cash flow. Free cash flow turned positive at $1.53 billion against a market expectation of negative $4.99 billion. Capex was $4.73 billion, up 110% but far below the $7.95 billion expected, and full-year guidance was cut from $25 billion to $20 billion. Musk is moving the battlefield from red-ocean car selling to the AI and space fronts where he still holds absolute advantage.
机器人被推到了台前。马斯克说Optimus已在加州弗里蒙特工厂进入"早期生产"，设计上还要调，达到每周2万台大概要一年或更长；瓶颈在训练数据，目标是明年4月前把AI5芯片塞进机器人。Cybercab年底可能开放外部购买，无监督FSD争取2027年中向部分客户放开。
Robots were pushed to center stage. Musk said Optimus has entered "early production" at the Fremont plant, still needs design tweaks, and reaching 20,000 units a week will take a year or longer; the bottleneck is training data, with AI5 chips targeted for the robot by April next year. Cybercab may open to outside buyers by year-end, and unsupervised FSD aims for some customers by mid-2027.
火箭公司的故事更性感。财报前一天，马斯克在SpaceX电话会上确认两家董事会已各自成立特别委员会，评估合并，他列了五大协同：超级智能软件、芯片设计、Terafab、机器人、星链。SpaceX刚交出营收增长161%、净利44亿美元的季报，摩根大通把目标价从240美元翻倍到500美元。合并像在雷区跳踢踏舞，成不成还得股东半数点头。
The rocket story is sexier. A day before the earnings, Musk confirmed on SpaceX's call that both boards have formed special committees to evaluate a merger, listing five synergies: superintelligence software, chip design, Terafab, robotics, and Starlink. SpaceX just posted revenue up 161% and net income $4.4 billion; JPMorgan doubled its target from $240 to $500. The merger is like tap-dancing in a minefield—it still needs a shareholder majority.
汽车的护城河还剩多深，是绕不开的老问题。比亚迪前三季度纯电累计领先特斯拉超过30万辆，小米YU7在国内细分市场几次压过Model Y。每当卖车被质疑，马斯克就抛一个更大的故事：欧洲销量崩了讲Robotaxi，交付首降讲Optimus量产，利润腰斩讲合并火箭。下一季，市场要听的不只是故事，是Robotaxi和Optimus真能送来钱。
Storage is another underrated bright spot. Q3 deployments hit 13.7 GWh, up 10% year on year, with record order growth and backlog; Megapack's 2027 capacity is nearly fully booked. The cash the cars no longer earn, Musk hopes storage and robots will slowly make up. He also said he is pushing for FSD approval in China—a huge market, but the pace sits in regulators' hands.
储能是另一个被低估的亮点。三季度储能装机13.7吉瓦时，同比增长10%，订单增速和积压量都创了新高，Megapack 2027年的产能几乎被预订一空。卖车赚不到的钱，马斯克指望储能和机器人慢慢补上。他还提到在努力争取FSD在中国获批，中国是个巨大市场，但落地节奏卡在监管手里。
How deep the car moat still runs is the unavoidable old question. BYD's front-three-quarter BEV lead over Tesla exceeds 300,000 units, and Xiaomi's YU7 has beaten Model Y in domestic segments several times. Each time car sales are doubted, Musk throws a bigger story: Robotaxi when Europe collapsed, Optimus mass production when deliveries first fell, a rocket merger when profit halved. Next quarter the market wants not just stories but proof that Robotaxi and Optimus actually send cash.
