Tesla Autopilot Fatal-Crash Lawsuits, Trials & Settlements
A string of high‑profile lawsuits over fatal crashes involving Tesla's Autopilot has produced jury verdicts, appeals and confidential settlements in 2025: a Miami federal jury found Tesla partly liable for a 2019 Key Largo crash and assigned roughly $242.5–$243 million to Tesla (with some outlets reporting a $329M total award), plaintiffs introduced data recovered from the vehicle after Tesla initially said it lacked a crash 'snapshot,' Tesla has filed appeals asking judges to overturn or retry verdicts, and the company quietly reached confidential settlements in at least two 2019 California Autopilot‑related death suits in mid‑September 2025. (reuters.com)
These developments mark a legal inflection point for Tesla and the broader autonomous‑vehicle/AI safety debate: juries are willing to assign corporate liability for harms tied to driver‑assist systems, plaintiffs are using forensic recovery (including third‑party/hacker‑recovered) vehicle data to challenge company narratives, settlements and punitive damage awards raise financial and disclosure risks for Tesla, and the outcomes are shaping regulatory, investor and policy discussions about how partially automated driving systems should be marketed, governed and limited. (washingtonpost.com)
Key parties include Tesla Inc. (defendant and Autopilot developer), plaintiffs and families from the 2019 crashes (notably the family of Naibel Benavides and survivor Dillon Angulo in the Florida case), plaintiff counsel (e.g., Singleton Schreiber / Brett Schreiber), defense counsel (including Gibson Dunn on appeal), federal judges in the Southern District of Florida, and investigative/regulatory observers and media (Reuters, CNBC, Washington Post) who have widely covered verdicts, appeals and settlements. (singletonschreiber.com)
- A Miami federal jury in early August 2025 found Tesla partially liable for a 2019 Key Largo crash and a jury award resulted in roughly $242.5–$243 million assigned to Tesla; some outlets reported a $329 million total award with Tesla’s share at about $242.5M. (reuters.com)
- Plaintiffs' experts recovered a crucial 'collision snapshot' and other telemetry that plaintiffs say Tesla had but initially told the court it did not, a development that plaintiffs’ teams say materially affected trial evidence and jury perception. (washingtonpost.com)
- Tesla has filed motions to toss the verdict or obtain a new trial and is disputing the size and legal basis of punitive damages; meanwhile, Tesla reached confidential settlements in at least two separate 2019 California Autopilot‑related death lawsuits in mid‑September 2025. (cnbc.com)
Regulatory Probes into Full Self-Driving (FSD) and Autopilot Safety
U.S. regulators have opened a preliminary evaluation into Tesla’s Full Self-Driving (FSD / “FSD Supervised”) software after dozens of reports alleging the system ran red lights, made illegal lane changes (including driving on the wrong side of the road), and in some cases failed to stop at railroad crossings; the investigation covers roughly 2.9 million Tesla vehicles and combines media reports, driver complaints and SGO crash reports gathered by NHTSA’s Office of Defects Investigation. (feeds.bbci.co.uk)
This probe escalates federal scrutiny of Tesla’s advanced driver assistance stack at a time when the company is testing robotaxi capabilities and marketing FSD to consumers; outcomes could range from mandated software fixes to recalls, restrictions on where or how FSD may be used, heightened congressional oversight, and a broader push for clearer rules and certification pathways for AI-driven vehicle systems. (techcrunch.com)
Primary actors are Tesla (developer/operator of FSD and the subject of complaints), the National Highway Traffic Safety Administration (NHTSA) and its Office of Defects Investigation (ODI) which opened the preliminary evaluation, and U.S. senators (notably Ed Markey and Richard Blumenthal) who have publicly pressed for investigations into FSD’s handling of railroad crossings; major news organizations (BBC, TechCrunch, Reuters, NBC, The Guardian) have documented the incidents and complaints shaping the probe. (techcrunch.com)
- NHTSA’s preliminary evaluation covers an estimated ~2.9 million Tesla vehicles equipped with FSD and cites dozens of reports (BBC reported 58 incidents referenced in the filing). (feeds.bbci.co.uk)
- NHTSA’s ODI said it received at least 18 complaints and identified multiple SGO crash reports (6 SGO reports cited by the agency) as part of the record that triggered the evaluation; the agency is examining scope, frequency and safety consequences and a Preliminary Evaluation can lead to further steps including an engineering analysis or recall. (techcrunch.com)
- Senators Ed Markey and Richard Blumenthal publicly urged NHTSA to probe FSD responses to railroad crossings on Sept 29, 2025, warning that failures at crossings could be 'catastrophic'—a line pulled directly from their letter. (blumenthal.senate.gov)
Model Y / 2021 Door-Handle Failures and US Safety Investigation
U.S. safety regulators (NHTSA) opened a preliminary investigation in mid-September 2025 into the electronic exterior door-handle system on approximately 174,000 Tesla Model Y vehicles from the 2021 model year after consumer reports that exterior handles sometimes become inoperative — including nine incident reports to regulators and four cases where parents said they had to break a window to regain access. NHTSA’s early review pointed to insufficient voltage in the vehicle’s low-voltage (12 V) electrical system as a possible cause, and Tesla’s design team has said it is working on a redesign that would combine the electronic and manual release mechanisms to make emergency egress more intuitive. (reuters.com)
The probe matters because it concerns a core safety function — door egress/egress during power loss or crashes — and could lead to a large recall, changes in vehicle hardware/software fail‑safes, new regulatory guidance on concealed/flush electronic handles, and heightened scrutiny of Tesla’s design and reporting practices; it also feeds broader debates about how software-defined vehicle functions (and OTA updates) interact with physical/hardware safety and emergency rescue procedures. (reuters.com)
Key actors include Tesla (vehicle manufacturer and Franz von Holzhausen, Tesla’s design chief, who has publicly discussed handle redesign options), the U.S. National Highway Traffic Safety Administration (NHTSA) as the investigator, journalists/outlets that raised and amplified the problem (Bloomberg, Reuters, New York Times, Ars Technica, TechCrunch), affected owners/parents who reported entrapment incidents, and safety/regulatory bodies in other jurisdictions watching design/regulatory precedents. (techcrunch.com)
- NHTSA opened a preliminary evaluation on September 16–17, 2025 covering roughly 174,000 Model Y (model year 2021) vehicles. (reuters.com)
- NHTSA reported nine consumer incident narratives (including four where parents said they broke a window to regain access); the agency’s initial review flagged low-voltage (12 V) supply to the electronic locks as a possible factor. (reuters.com)
- Tesla’s design chief said the company is considering combining the electronic actuation and manual release into a single, more intuitive control as part of a redesign in response to the scrutiny. (techcrunch.com)
Cybertruck Crashes, Fatalities, Lawsuits and Repurposing
Since late 2024 a string of high-profile incidents and safety actions has converged around Tesla's Cybertruck: a high‑speed crash on November 27, 2024 in Piedmont, California that produced three fatalities and left families alleging passengers were trapped by the Cybertruck's electrically‑actuated door system has led to multiple wrongful‑death lawsuits filed in early October 2025; federal safety regulators have opened probes into electronic door failures on Tesla vehicles; at the same time the Cybertruck is experiencing a steep sales slump and repeated recalls, prompting Tesla to move unsold units into affiliated fleets (SpaceX and xAI) and drawing unconventional reuse — the U.S. Air Force has requested Cybertrucks as target vehicles for precision‑munitions testing — all of which has amplified legal, regulatory and reputational pressure on Tesla. (apnews.com)
This matters because it ties vehicle design, safety oversight, corporate risk and even national‑security training into a single trend: alleged design shortcomings (electronic doors, hard‑to‑access manual releases) are being litigated after deadly crashes, regulators (NHTSA) are investigating, widespread recalls and sharply lower sales strain Tesla's business, Tesla's internal purchases/repurposing to SpaceX/xAI raise questions about how manufacturers manage slow‑moving inventory, and military interest in Cybertrucks as targets highlights dual‑use concerns and public perception problems — together these developments could influence regulatory action, civil liability exposure, consumer trust, and how AI‑led features (Autopilot/FSD) and electronic systems are engineered and audited. (reuters.com)
Principal actors are Tesla (vehicle maker and defendant), CEO Elon Musk (public face and owner of related companies), the families and attorneys filing wrongful‑death suits (including the Tsukahara and Nelson families), federal regulators (NHTSA, which has opened related probes), the U.S. Air Force / Air Force Test Center (contracting Cybertrucks for target testing), allied Musk companies receiving inventory (SpaceX and xAI), and the Alameda County Superior Court where suits were filed — media, safety researchers and insurers/IIHS are also active observers. (techspot.com)
- Nov. 27, 2024 — a Cybertruck crash in Piedmont, California killed three occupants (driver Soren Dixon and passengers Jack Nelson and Krysta Tsukahara); plaintiffs say rear passengers survived the impact but died of smoke inhalation after being unable to open electronic doors; families filed amended wrongful‑death suits in Alameda County in early October 2025. (apnews.com)
- Aug. 2025 — U.S. Air Force contracting documents requested two Tesla Cybertrucks (among 33 target vehicles) for Stand Off Precision Guided Munitions testing at White Sands, citing the truck's unique stainless‑steel exoskeleton and 48V architecture — effectively repurposing unsold Cybertrucks as missile/munitions targets. (stripes.com)
- Tesla has been moving large batches of unsold Cybertrucks into internal fleets (SpaceX and xAI) as sales slumped: Tesla reported roughly 4,306 Cybertrucks sold in Q2 2025 and about 5,385 in Q3 2025 (≈16,000 YTD 2025), far below original targets; the model has undergone multiple recalls (eight reported) since launch. (techspot.com)
Dojo Supercomputer Shutdown, Team Disbanding & Executive Exits
In early August 2025 Tesla ordered the shutdown and disbanding of its Dojo supercomputer project: the Dojo team was dissolved, its leader Pete (Peter) Bannon is exiting Tesla, roughly ~20 engineers from the group moved to a new startup called DensityAI, and remaining staff were reportedly reassigned to other compute/data‑center efforts as Tesla pivots to rely more on external partners and its AI5/AI6 chip road‑map. (news.bloomberglaw.com)
The move marks a material strategic shift for Tesla’s in‑house AI infrastructure ambitions — abandoning a years‑promoted, vertically integrated training supercomputer approach (Dojo/D1→D2) in favor of concentrating on inference chips (AI5/AI6), third‑party GPUs (Nvidia/AMD) for training, and contracting Samsung to fabricate next‑gen chips; it affects Tesla’s autonomy roadmap, talent retention, potential new markets (Dojo-as‑service), and competitive positioning in AI/semiconductor supply chains. (techcrunch.com)
Key actors are Tesla (CEO Elon Musk; Dojo lead Pete Bannon and other Dojo engineers), the newly formed DensityAI (founded/led by ex‑Dojo personnel including Ganesh Venkataramanan and others), chip/foundry partners Nvidia and AMD (compute suppliers), and Samsung (the reported $16.5B AI6 manufacturing deal). Major reporters covering the development: Bloomberg, Reuters, TechCrunch, CNBC and TechSpot. (news.bloomberglaw.com)
- Dojo team disbanded and project shut down in early August 2025; Pete (Peter) Bannon — the Dojo lead — is leaving Tesla. (news.bloomberglaw.com)
- About ~20 former Dojo engineers left Tesla and have joined (or are reported to have joined) a new data‑center startup, DensityAI; remaining Dojo staff were reassigned inside Tesla to other compute and data center projects. (news.bloomberglaw.com)
- Elon Musk framed the decision as a resource/convergence choice — saying it made little sense to run two divergent chip architectures and that Tesla will focus effort on AI5/AI6 inference chips while relying more on Nvidia/AMD for training compute. (Paraphrase of Musk’s statements and posts). (techspot.com)
Optimus Robotics Program, Rival Robots & AI Talent Exodus
Throughout 2025 Tesla’s high‑profile Optimus humanoid program has entered a period of competing signals: Elon Musk is publicly betting that Optimus will be a core value driver for Tesla (he posted in early September 2025 that “~80% of Tesla’s value will be Optimus”), and the company has set aggressive volume targets (Musk has referenced “thousands” of robots in 2025 and 1 million/year by 2030; some reporting cited targets of ~5,000 in 2025 and 50,000 in 2026), while at the same time senior robotics and AI leaders have left (Optimus program head Milan Kovac resigned on June 6, 2025; an Optimus AI lead, Ashish Kumar, announced his move to Meta on September 19, 2025) and Tesla has scaled back in‑house compute (Dojo team dissolution in August 2025) — coinciding with rising competition from other robot makers (Unitree and X‑Humanoid won multiple medals at the inaugural World Humanoid Robot Games in Beijing, Aug 17–18, 2025).
This matters because Tesla’s narrative and investor valuation increasingly depend on a future in which physical AI (robotaxis + humanoids) delivers the bulk of growth; leadership departures, Dojo’s shutdown, supply constraints (reported China rare‑earth export limits), and stronger rivals in China/elsewhere introduce technical, execution and talent risks that could delay or reduce the size of the payoff Musk projects — while competitors and deep‑pocketed AI players (Meta, Apple, Google/OpenAI) are actively hiring the same AI/robotics talent.
Key organizations and people are Tesla (CEO Elon Musk; former Optimus head Milan Kovac; Autopilot lead Ashok Elluswamy who was named to take on Kovac’s responsibilities; Optimus AI lead Ashish Kumar who moved to Meta), Meta (hiring AI talent from Tesla), Chinese robot makers Unitree and X‑Humanoid (strong showings at Aug 2025 World Humanoid Robot Games), xAI/X (part of the broader Musk talent and political controversy), and major infrastructure partners Nvidia/AMD/Samsung (Tesla moving away from Dojo toward external compute partnerships).
- Unitree and other Chinese firms dominated the inaugural World Humanoid Robot Games in Beijing (Aug 17–18, 2025): the event featured ~280 teams from 16 countries and Unitree won 11 medals including 4 golds, underlining rising competitive capability outside Tesla.
- Leadership and talent shifts in 2025: Milan Kovac (head of Optimus) resigned June 6, 2025; Tesla dissolved or reassigned its Dojo supercomputer team in early August 2025; Ashish Kumar, identified as an Optimus AI lead, announced leaving Tesla for Meta on September 19, 2025 — part of reporting about a broader senior exodus around late Sep 2025.
- Elon Musk’s public position: he has pushed an optimistic long‑term thesis for Optimus (telling investors / posting in early September 2025 that ~80% of Tesla’s future value could stem from Optimus) even as reporters and analysts flag execution, talent and supply‑chain headwinds.
Andrej Karpathy, nanochat Release & AI Commentary
Andrej Karpathy publicly released nanochat on October 13, 2025 — an open-source, dependency-light, full‑stack ChatGPT‑style training + inference pipeline (roughly ~8,000 lines of code) that includes a Rust BPE tokenizer, pretraining on FineWeb‑EDU, mid‑training/SFT, an optional tiny RL loop, a thin inference engine with KV cache and a simple web UI; Karpathy says the repo can produce a usable chat model on a rented 8×H100 node in about 4 hours for roughly $100, with larger runs (e.g., ~33 hours / ~$1,000) producing measurably stronger micro‑models. (simonwillison.net)
Nanochat matters because it (a) makes an end‑to‑end LLM training/inference workflow readable, hackable and reproducible for students and developers, (b) sets concrete, low‑cost GPU speedrun benchmarks that democratize experimentation (8×H100 ≈ 4 hours / ≈ $100), and (c) reframes discussions about agentic AI by emphasizing the long, incremental engineering work still required — Karpathy has used the release and his interviews to argue for measured expectations about agents and better curated data rather than hype. (simonwillison.net)
Primary actor: Andrej Karpathy (author of nanochat and founder of Eureka Labs; formerly Tesla's director of AI/Autopilot and an early OpenAI researcher). Secondary players and ecosystems implicated: the open‑source GitHub community (repo/discussions), cloud GPU vendors / NVIDIA (H100 hardware referenced for speedruns), and the broader AI industry debating 'agent' timelines (investors, startups, and companies like OpenAI/Tesla are relevant stakeholders). (reuters.com)
- Release date and repo scale: nanochat announced by Karpathy on October 13, 2025; codebase reported at ~8,000 lines of code. (simonwillison.net)
- Cost/time baseline: Karpathy provided a speedrun configuration that can train a usable chat model in ~4 hours on a rented 8×NVIDIA H100 node for roughly $100; larger runs (e.g., ~33 hours / ~$1,000) produce stronger micro‑models and measurable metric gains. (simonwillison.net)
- Key positions/quotes: Karpathy said the nanochat repo is 'basically entirely hand‑written (with tab autocomplete)' and has argued publicly that current agentic AI is overhyped and 'it will take about a decade' to work through core issues (memory, multimodality, reliable tool use). (analyticsindiamag.com)
Shanghai Gigafactory Ramp-up & High-Automation Claims
Tesla’s Shanghai Gigafactory has officially entered a fourth‑quarter 2025 production ramp-up as announced by Tesla China vice‑president Tao Lin (Weibo / Reuters), while independent/specialist sites and social videos circulating in early October have claimed that a new Model YL line in Shanghai runs at extremely high automation levels (reports allege up to ~95% robotic steps, 2.5‑hour "construction" cycle and production speeds several times faster than prior lines). (reuters.com)
If the ramp‑up is paired with materially higher automation and AI‑driven quality/line management, it affects Tesla’s global output, export mix from Shanghai, labor composition on the line, and the broader manufacturing race toward "lights‑out" facilities — with implications for unit costs, time‑to‑market for new variants, supply‑chain throughput and EV competition in China and export markets. However, some high‑automation claims remain unverified by major independent outlets. (reuters.com)
Key actors are Tesla (Shanghai Gigafactory operations, corporate leadership including Tesla China VP Tao Lin and global management), local Shanghai authorities and reporting that highlight Shanghai output and exports, and independent commentators / technology blogs (notably NextBigFuture / Brian Wang) and social‑media sources that circulated factory footage and automation claims. Equipment/robotics and AI vendors (not named in the sensational pieces) would be consequential if claims are accurate. (reuters.com)
- Reuters reported that Tesla’s Shanghai Gigafactory began a Q4 2025 production ramp‑up announced by Tesla China VP Tao Lin on October 13, 2025. (reuters.com)
- NextBigFuture and circulated factory videos claim a Model YL production line in Shanghai is ~95% robotic, completes certain vehicle "construction" steps in ~2.5 hours and operates up to ~4x faster than earlier Tesla lines (articles published Oct 6–8, 2025). (nextbigfuture.com)
- Official local reporting and Shanghai government summaries cite large output and export volumes from the Shanghai complex (high quarterly deliveries and significant export share), but do not independently confirm the full 95% automation metric in mainstream international outlets. (english.shanghai.gov.cn)
Sales, Deliveries, Earnings & Competitive Pressure (BYD, EU Slump)
Tesla’s near-term sales picture has been volatile in 2025: the company reported a sharp second‑quarter auto revenue decline and related investor pain in July, then posted a record third‑quarter production and deliveries of 497,099 vehicles on October 2, 2025 as buyers rushed to capture expiring U.S. EV tax credits — but that delivery bounce has not erased structural headwinds including an ongoing sales slump in Europe and intensifying competition from China’s BYD, which has outsold Tesla on a quarterly BEV basis for four straight quarters. (cnbc.com)
This matters because (1) short‑term incentives (like the $7,500 U.S. credit) can temporarily inflate volumes and margins but mask deteriorating underlying demand in key markets (Europe, parts of China); (2) BYD’s vertical integration, product cadence and low‑cost models are eroding Tesla’s unit and market share globally; and (3) investor narratives are increasingly driven by Tesla’s AI/autonomy ambitions (FSD, robotaxi, Optimus, Dojo) — a story with huge upside if realized but that currently contributes little to revenue, leaving a valuation/expectations gap and heightened sensitivity to quarterly sales and earnings. (reuters.com)
Primary companies and actors are Tesla Inc. (Elon Musk, management/IR), BYD Co. (Wang Chuanfu/management), major financial analysts and banks (e.g., Barclays, BNP Paribas cited in coverage), market data providers and regulators (ACEA for Europe registration data; NHTSA and other safety/regulatory bodies in the U.S.), and news/market outlets that move sentiment (CNBC, Bloomberg, Reuters, Tesla Investor Relations). (ir.tesla.com)
- Tesla reported Q3 2025 deliveries of 497,099 vehicles (record quarterly deliveries) on Oct 2, 2025. (ir.tesla.com)
- Tesla’s auto revenue fell to $16.7 billion in Q2 2025 (year‑over‑year decline) and the company warned of possible "a few rough quarters," triggering an ~8% intraday share drop after the report in late July 2025. (cnbc.com)
- Analysts and some banks highlight a sharp valuation debate: several sell‑side notes (e.g., BNP Paribas / exane coverage cited in analyst summaries) argue Tesla’s robotaxi/Optimus/AI narrative underpins a large portion of its market value today despite those businesses producing little or no revenue. (teslarati.com)
Tesla Insurance Arm Regulatory Scrutiny & Mishandled Claims Allegations
California’s Department of Insurance (CDI) filed formal enforcement actions in early October 2025 accusing Tesla Insurance Services, Tesla Insurance Company and partner underwriter State National of widespread mishandling of auto insurance claims — alleging “egregious delays,” unreasonable denials, inadequate investigations and systemic failures that produced nearly 3,000 violations of California insurance law since 2022 and hundreds-to-thousands of consumer complaints; the filings give the companies 15 days to respond and threaten hearings, fines and possible suspension or revocation of their California insurance licenses. (insurance.ca.gov)
This matters because the CDI’s action can impose sizable civil penalties (up to $5,000 per unlawful act and up to $10,000 per willful violation), put Tesla’s in-house insurance business at risk of losing the ability to operate in the nation’s largest auto-insurance market, accelerate related private litigation (including a July 2025 proposed class action), and intensify scrutiny of how automakers — particularly those that rely on vehicle telematics and automation-related data — manage claims and use automated tools; the case also amplifies policy debates about insurer use of automated decisioning and telematics under existing and emerging California rules. (insurance.ca.gov)
The principal actors are the California Department of Insurance (Commissioner Ricardo Lara) as regulator and plaintiff in the enforcement filings; Tesla Insurance Services, Inc. and Tesla Insurance Company (the insurer/producer units owned by Tesla, Inc.); State National Insurance Company (the admitted carrier/underwriter named in the filings); California policyholders and consumer-plaintiff attorneys (including the proposed class action filed July 11, 2025); and secondary stakeholders such as other state regulators and consumer-protection groups following insurer automation and telematics practices. (insurance.ca.gov)
- CDI’s filings say consumer complaints and identified violations surged in 2025: the regulator reported figures that amount to roughly 1,481 complaints and 1,969 identified violations against the Tesla Companies through September 22, 2025, and characterized the cumulative total since 2022 as nearly 3,000 violations. (techcrunch.com)
- The Department of Insurance gave Tesla Insurance and State National 15 days to respond to formal 'Accusations' and Notices of Order to Show Cause (filed Oct 3–4, 2025); absent pre-hearing resolution, CDI will seek an administrative hearing that could suspend or revoke their California licenses and impose fines up to $5,000 per violation or $10,000 per willful violation. (insurance.ca.gov)
- CDI’s public assertion: the companies 'placed profits above people' — a central formulation the regulator used to characterize the alleged pattern of willful unfair claims settlement practices. (insurance.ca.gov)
FSD Consumer Perception, Adoption Surveys & Feature Updates
Over the past two months public reaction and regulatory scrutiny around Tesla's Full Self-Driving (FSD) has intensified: an August 2025 Slingshot Strategies / Electric Vehicle Intelligence survey of ~8,000 Americans found only 14% said FSD would make them more likely to buy a Tesla while 35% said it would make them less likely — and nearly half said FSD should be illegal — at the same time Tesla pushed a major driver‑assist software rollout (FSD v14) to a limited group on October 7, 2025 and U.S. regulators (NHTSA) opened a probe on October 9, 2025 covering roughly 2.88 million Teslas after reports of red‑light/traffic violations and several crashes. (cnbc.com)
This confluence — weak consumer sentiment toward FSD, an aggressive product update cadence from Tesla, and a formal NHTSA investigation — matters because it affects demand and brand reputation (potentially depressing EV sales), raises the prospect of regulatory actions or mandated fixes (including recalls), and alters the commercial timetable for robotaxi/ride‑hailing ambitions and monetization of autonomous services. It also shapes investor expectations for Tesla's software‑led revenue strategy and forces competitors and regulators to reassess safety and disclosure standards for driver‑assist systems. (cnbc.com)
Tesla and CEO Elon Musk (product release and messaging), Slingshot Strategies / Electric Vehicle Intelligence (survey publisher), NHTSA (U.S. safety regulator leading the probe), media outlets covering incidents and adoption (CNBC/Reuters/Washington Post), and competing self‑driving projects (Waymo, Baidu/Apollo) and other automakers — all of whom influence public perception, regulatory posture, and competitive benchmarks. (cnbc.com)
- Slingshot Strategies' Electric Vehicle Intelligence survey (Aug 2025) polled ~8,000 U.S. adults and found 14% would be more likely to buy a Tesla because of FSD while 35% would be less likely; 51% said it made no difference and ~48% said FSD should be illegal. (cnbc.com)
- Tesla rolled out a major driver‑assist / FSD software update (reported as FSD v14) to a limited group on October 7, 2025 as it continues iterative releases ahead of broader publicity/unveils. (investors.com)
- The U.S. National Highway Traffic Safety Administration opened a preliminary investigation announced October 9, 2025 into roughly 2.88 million Tesla vehicles equipped with FSD following reports of traffic violations (including red‑light events) and multiple crashes / injuries. (reuters.com)
Executive Departures & Talent Exodus Across Tesla and xAI
Throughout August–September 2025 a wave of senior departures and team shutdowns has hit Tesla and Elon Musk’s AI venture ecosystem: Tesla disbanded its in‑house Dojo supercomputer team in early August and its Dojo lead Peter (Pete) Bannon departed (reporting around Aug 7, 2025), roughly ~20 engineers left prior to the shutdown to form a new startup called DensityAI, and in mid‑to‑late September Tesla’s lead for Optimus AI, Ashish Kumar, left Tesla to join Meta — all occurring as Tesla shifts toward external partners (Nvidia/AMD/Samsung) and a $16.5 billion Samsung chip/manufacturing deal. (reuters.com)
This matters because the moves represent both a strategic pivot (Tesla stepping back from bespoke, large‑scale in‑house training infrastructure like Dojo in favor of third‑party compute and a consolidated chip roadmap) and a talent risk: the loss of specialized engineers and leaders can slow Optimus/humanoid and autonomy timelines, reshape competitive dynamics for AI talent (Meta, OpenAI and others are actively hiring), and raise governance/operational questions about morale and recruitment under Musk’s management style. (techcrunch.com)
Primary organizations and people include Tesla (CEO Elon Musk; Dojo lead Peter/Pete Bannon; Optimus AI lead Ashish Kumar; other departed executives such as Milan Kovac and David Lau), xAI/X (organizationally linked to Musk and implicated in related turnover), DensityAI (the new startup formed by ~20 former Dojo engineers), and competitors/partners including Meta (which hired Kumar), Nvidia, AMD and Samsung (partnered for chips). (cnbc.com)
- Around Aug 7, 2025: reporting that Tesla ordered the Dojo supercomputer team to be disbanded and that Dojo lead Peter (Pete) Bannon was leaving the company. (cnbc.com)
- Approximately ~20 engineers from Tesla’s Dojo group departed to start a new data‑center/AI infrastructure startup called DensityAI before the Dojo disbanding was announced. (reuters.com)
- Key quote from an insider/leader: Ashish Kumar (Optimus AI lead) on X: 'Decided to leave Tesla. It’s been an incredible ride leading the Optimus AI team… AI is the most significant bit to unlock humanoids.' (tweet announcing his move to Meta, Sept 19, 2025). (teslarati.com)