July 23, 2026 · Thursday2026 年 7 月 23 日 · No. 19 · updated更新于 11:11 () SIGNALS信号EVENTS日历ARCHIVE归档ACCESS接入ABOUT关于

AUTOSIGNAL

Human-curated. Expert-annotated. Every signal traced to source. 人工精选 · 专家点评 · 每条信号可溯源

← Events← 返回日历

EVENT BRIEF · 事件摘要与解读

Tesla Q2 2026 Earnings Call特斯拉 2026 年二季度财报电话会

2026-07-22 Earnings · 财报 Brief v1 · 电话会当日

OFFICIAL MATERIALS官方材料直达

KEY NUMBERS · SOURCE: SEC FILING, NOT MEDIA核心数字 · 取自 SEC 备案而非媒体转述

Metric指标Q2 2026本季Read读法
Revenue营收$28.24B (+26%)Beat; first >$100B trailing-12M超预期;滚动年营收首破千亿
Operating income经营利润$398M · 1.4% (−57%)The real story of the quarter本季真实底色
GAAP net incomeGAAP 净利$1,114MIncl. +$1,005M SpaceX unrealized gain +$274M tax items; core ≈ negative含 SpaceX 未实现收益 +$10.05 亿与税项 +$2.74 亿;剔除后核心净利约为负
Auto GM ex-credits汽车毛利率(剔积分)16.3% (+130bp YoY)Kills the "discount-bought volume" bear case证伪「降价换量」空头叙事;miss 来自费用端
Capex / FCF$5.79B (+142%) / −$1.09B2nd straight negative FCF quarter连续第二季 FCF 为负
Deliveries交付480,126 (Q2 record)Known since Jul 27 月 2 日已公布
FSD1.48M subs (+56%), >55% attach订阅 148 万(+56%),北美新车附着率>55%Genuine monetization datapoint软件变现的真实进展
Robotaxi7 metros; ~2.5M cum. paid miles7 城;累计付费里程约 250 万英里See our triangulation below规模详见下方解读③
Energy storage储能13.5 GWh (+41%)Revenue only +13% — price deflation收入仅 +13%,单价通缩
After-hours盘后股价−3 ~ −4.7%Options priced ±7.6%期权定价 ±7.6%

MARKET CONSENSUS IN ONE PARAGRAPH市场共识一段话

Street narratives: record revenue vs vanishing profit; the first real "AI capex test"; robotaxi city-count land-grab vs Waymo's operating reality; the SpaceX question. The $130–$600 target spread (Wells Fargo ↔ Wedbush) is almost entirely a disagreement about autonomy option value — not the car business.媒体四大叙事:营收纪录 vs 利润消失、AI 资本开支大考、Robotaxi 攻城 vs Waymo 现实、SpaceX 合并悬念。$130–$600 的目标价极差(Wells Fargo ↔ Wedbush)几乎全部来自对 autonomy 期权价值的分歧——与卖车业务无关。分析师白空:robotaxi 单位经济性、车队缩水原因、Cybercab 硬日期、capex 构成均无人能答。

AUTOSIGNAL READ · ANALYSIS PILOTAUTOSIGNAL 解读 · 分析尝试

Editorial analysis pilot built on primary data (NHTSA SGO filings, SEC documents, quarterly-deck text analysis, our own signal archive). Assumptions flagged inline. Not investment advice.基于一手数据(NHTSA SGO 备案、SEC 文件、股东信文本定量分析、自有信号档案)的编辑部分析尝试;推算假设均已标注。不构成投资建议。

Profit quality利润质量注记
The $1.1B GAAP headline is flattered by a $1.0B non-cash SpaceX gain and $274M of tax items — core net income was roughly negative. At 1.4% operating margin, Tesla now runs AI-infrastructure capex on automaker margins. 「GAAP 净利 $11 亿」被 SpaceX 非现金收益与税项两笔粉饰,剔除后核心净利约为负。1.4% 的经营利润率说明:特斯拉已经是一家「用汽车利润率支撑 AI 基建开支」的公司。
Regulator-filed data vs deck language监管备案数据 vs 官方叙事
NHTSA SGO filings (our own processor, primary CSVs): Tesla consumer Autopilot/FSD incident reports rose 152% in 10 months (82→207/mo, May record) and now represent 94% of all industry L2 reports. In the same quarter, "safety" mentions in the shareholder deck dropped 5→5→1. The narrative exits safety precisely as regulatory heat peaks (France FSD pause, NJ sensor bill, NHTSA warnings — all this month). NHTSA SGO 备案(自建处理器,一手 CSV):特斯拉消费级 Autopilot/FSD 月度事故报告 10 个月上升 152%(82→207 起,5 月创纪录),占全行业 L2 报告的 94%。而同期股东信中「safety」词频 5→5→1——在监管热度最高的月份(法国暂缓 FSD、新泽西传感器法案、NHTSA 警告齐发),官方叙事恰好退出了安全话题。
Robotaxi scale triangulationRobotaxi 规模三角测量
Two independent methods agree: regulator filings (Tesla 22 cumulative ADS reports vs Waymo 1,013 in the same period) and paid-miles math (~140K mi/wk vs Waymo ~2.5M) both put Tesla's actual robotaxi operating scale at ≈5–6% of Waymo's — while headlines count cities (7 vs 10). Incident rates per million miles are comparable (~7–9), so the gap is scale, not safety performance. [Assumes ~5 mi/ride for Waymo; ±30% doesn't change the order of magnitude.] 两条独立路径同指一个结论:监管备案(Tesla ADS 累计 22 起 vs Waymo 同期 1,013 起)与付费里程折算(约 14 万英里/周 vs Waymo 约 250 万)都表明,Tesla robotaxi 的实际运营规模约为 Waymo 的 5–6%——而媒体标题在数城市(7 vs 10)。每百万英里事故率两家相近(约 7–9 起),差距在规模而非安全表现。[假设 Waymo 单程约 5 英里;±30% 误差不改变量级]
Deck language drift (Q4'25→Q2'26)股东信语言漂移(三季定量对比)
AI mentions fell 19→12 while AI capex rose 142% (talk less, spend more); "affordable" went to zero (that narrative is over); "unsupervised" 3→9 and Cybercab 4→8 (the chips keep moving to autonomy) — yet Cybercab language stays carefully unfalsifiable ("production commenced"). 「AI」词频 19→12 而 AI 资本开支 +142%(说得更少,花得更多);「affordable」归零(平价车型叙事正式结束);「unsupervised」3→9、「Cybercab」4→8(筹码持续押向自动驾驶)——但 Cybercab 措辞仍是不可证伪的「production commenced」。
Announcement-calendar coupling扩张公告与财报日历的耦合
In the 20-day earnings window our archive logged 5 negative regulatory events vs 2 expansion announcements — and Orlando/Tampa was announced 24 hours before this call, the second straight expansion timed to a financial event. Pattern to re-test at Q3: watch Phoenix/Las Vegas timing. 财报前 20 天窗口,自有档案记录监管负面 5 起 vs 扩张公告 2 起;奥兰多/坦帕在电话会前 24 小时官宣,连续第二次踩点财务事件。「监管密集期的叙事对冲」模式成立与否,Q3 财报前看凤凰城/拉斯维加斯的官宣时点即可第三次验证。

THE BOTTOM LINE一句话结论

The market is arguing about what the autonomy option is worth; the primary data says the cost of exercising it is rising — regulatory, operational and narrative indicators all diverged from the option-value debate this quarter, and neither bulls nor bears have priced that.市场在争论 autonomy 期权值多少钱;一手数据说的是行权成本正在上升——监管、运营、叙事三个维度本季同时与期权价值之争背离,而多空双方都没有为此定价。

WATCHLIST (POST-CALL)后续观察清单

□ Fleet size disclosure vs our 5–6% estimate · □ A hard Cybercab SOP date (2027H1 admission?) · □ Capex split (compute/Optimus/Cybercab) · □ Management response on France/EU approval · □ "Sell vs own fleet" allocation language (already seeded in the Outlook) · Transcript-based Q&A language analysis will be appended when transcripts post.□ 若披露车队数字→与 5–6% 推算交叉验证 · □ Cybercab 硬日期(承认 2027H1?) · □ 季度 $58 亿 capex 构成拆分 · □ 管理层对法国/欧盟审批的回应口径 · □「卖车 vs 自留车队」分配表态(Outlook 已埋伏笔) · 转写发布后将补问答语言分析。