Nuro

Phase A — Understand the business

Lens 1 · Company Overview

What it is. Nuro is a Mountain View self-driving-technology company founded September 2016 by two ex-Google/Waymo principals, Jiajun Zhu and Dave Ferguson. For its first eight years it was a last-mile delivery company: it designed, built, and operated its own small driverless cargo pods (R1 2018 → R2 2020 → R3/"Nuro" 3rd-gen 2023, a 500-lb-payload, 45-mph, battery-electric pod with an external pedestrian airbag) and ran grocery/food/pharmacy delivery pilots. In September 2024 it pivoted the entire business model: it stopped trying to own-and-operate a delivery fleet and instead began licensing its autonomy stack — "the Nuro Driver" — to automakers, tier-1 suppliers, and mobility platforms.

How it now makes money (the thesis). Two licensable products:

• Nuro Driver (SAE Level 4) — full driverless autonomy for robotaxis, commercial fleets, and eventually personal vehicles. Automotive-grade sensors + compute (NVIDIA DRIVE Thor / Hyperion) + Nuro's "AI-first" software, integrated into a partner's vehicle on the production line.
• Nuro Driver Assist (Level 2++) — an eyes-on ADAS product for personal cars, the near-term revenue bridge while L4 matures.
• Plus a Nuro AI Platform / Toolkit — validation and simulation tooling sold to third-party developers.

The revenue model is not disclosed — no per-mile fee, upfront license, or royalty rate is public. The structure is the classic "Intel Inside / arms-dealer" play: Nuro supplies the brain, the OEM supplies the car, the mobility platform (Uber) owns and operates the fleet and takes the customer relationship. Nuro carries no vehicle capex and no fleet-operations cost — the explicit rationale for the pivot.

Main relationships.

• Anchor customer / partner: Uber + Lucid. July 2025 deal to deploy 20,000+ Lucid Gravity SUVs running the Nuro Driver L4 over six years, Uber owning/operating; expanded April 2026 to ≥35,000 vehicles with Uber lifting its Lucid investment to $500M total. First engineering vehicles delivered Nov 2025; SF launch targeted 2026.
• Compute supplier: NVIDIA (DRIVE Thor / Blackwell, DriveOS) — also now an investor (Series E). Lenovo and Arm are named build partners.
• Legacy delivery partners (now dormant/de-emphasized): Kroger, Domino's, CVS, Chipotle, FedEx, 7-Eleven, Walmart, Uber Eats (a 10-yr 2022 food-delivery deal that predates and is orthogonal to the robotaxi pivot).

Contract structure & concentration. This is the whole story: revenue concentration is ~total in one commercial relationship (Uber/Lucid) that has not yet produced a single paid mile. Take-or-pay terms are undisclosed; the risk is that licensing revenue is back-loaded and contingent on Nuro clearing a driverless permit stack (CPUC + DMV) and Lucid/Uber actually scaling. See Lens 13.

Lens 2 · Supply Chain

Map, upstream → Nuro → end rider, naming every stakeholder:

Upstream (into the Nuro Driver):

• Compute: NVIDIA DRIVE AGX Thor (Blackwell-architecture SoC) + DriveOS + Hyperion reference platform. This is a single-source dependency on NVIDIA silicon — a chokepoint, and notable given NVIDIA is also a shareholder and is arming Nuro's competitors (Wayve, others) on the identical platform.
• Sensors: Nuro-specified automotive-grade lidar/radar/camera suite (specific tier-1s not publicly enumerated post-pivot; historically the pods used a custom sensor stack). LiDAR supply in the sector concentrates on Hesai/Luminar/Innoviz-class vendors — Nuro's exact vendor list is not disclosed [n/a — private, not disclosed].
• Systems integration hardware: Lenovo (compute integration), Arm (CPU IP in the compute path).
• The vehicle itself: Lucid builds the Gravity SUV and integrates the Nuro Driver on the production line (factory-fit, not retrofit).

Nuro (the value-add): the L4 software — perception, prediction, planning, an increasingly end-to-end/foundation-model driving policy — plus the simulation and validation toolchain. This is the only piece Nuro owns; everything on either side is someone else's balance sheet.

Downstream (Nuro → rider):

• Fleet owner/operator: Uber — owns the vehicles, runs depots, remote assistance, charging, and the rider app.
• End customer: the Uber robotaxi rider, initially in San Francisco (2026).

Chokepoints / single-source flags.

1. NVIDIA compute — sole compute platform; also a competitor-arming shareholder.
2. Lucid — sole vehicle for the anchor program; Lucid is itself a cash-burning, going-concern-adjacent EV maker (Uber's $500M lifeline is telling) — a fragile leg. If Lucid stumbles on Gravity production, Nuro's flagship deployment stalls through no fault of its own software.
3. Uber — sole route to the rider; Uber simultaneously partners with Waymo, WeRide, Pony.ai, Wayve and others. Uber is a frenemy aggregator, not a captive channel.

Names-or-it-didn't-happen check: NVIDIA, Lucid, Uber, Lenovo, Arm — all confirmed. Sensor tier-1s: not disclosed.

Lens 3 · Competitive Advantages (moats)

The honest moat inventory — thin but non-zero:

1. Founder DNA / talent. Zhu (Waymo perception + simulation lead) and Ferguson (Waymo CV/ML/prediction lead; DARPA Urban Challenge winner with CMU 2007) are genuinely top-tier AV pedigree; each holds 100+ patents. This is a real but depreciating moat — the entire sector is staffed by ex-Waymo/Cruise people.
2. Deployment track record & regulatory firsts. First-ever NHTSA autonomous exemption (2020); first California AV deployment permit (2020). Claimed 1.4–1.7M autonomous L4 miles, 5+ years, zero at-fault incidents. Regulatory relationships and a clean safety ledger are a modest switching/credibility moat with OEMs and permitting bodies.
3. Capital-light architecture (the pivot's whole point). By licensing rather than owning fleets, Nuro claims it can scale on other people's capex — a structurally higher-ROIC model if it lands multiple licensees. This is a business-model advantage, not a technology moat.
4. "Second-mover" learning. Ferguson's explicit thesis: let Waymo eat the pioneer's regulatory/PR costs, then arrive with an AI-first, end-to-end stack that's cheaper to generalize. Plausible narrative; unproven.

Bargaining power — weak on both sides. Nuro needs Uber/Lucid far more than they need Nuro (Uber has five AV partners; Lucid could integrate a rival stack). Upstream, it is a price-taker to NVIDIA. A licensor with one customer and one silicon supplier has negative bargaining leverage until it proves the stack and adds licensees. Verdict: the moat is a clean safety record + elite founders + a capital-light model — none of which is defensible against Waymo's data flywheel or Tesla's fleet-scale data. The durable-moat question (scale/network/switching/IP) mostly answers "not yet."

Lens 4 · Segments

+private overlay — no audited segment P&L exists. Segment breakdown is n/a — private, not disclosed for revenue/EBIT/geography. What can be stated structurally:

• Historic segment (2018–2024): last-mile delivery — B2B pilots priced below human delivery, revenue immaterial vs. spend. Now wound down as an operating business.
• Current segments (post-pivot, pre-revenue at scale):
• Nuro Driver L4 licensing — robotaxi (Uber/Lucid) is the entire pipeline today. Revenue: back-loaded, tied to 2026+ deployment.
• Nuro Driver Assist L2++ — personal-vehicle ADAS licensing; the intended near-term cash bridge; no signed OEM disclosed [n/a].
• Nuro AI Platform / Toolkit — validation software; nascent.
• Geography: US-centric (SF launch); testing expanded to Japan (Apr 2025) and Las Vegas Strip (May 2025). No revenue geography disclosed.

Trend: the company is mid-transition from a zero-revenue operator to a would-be licensor with one un-launched contract — there is no segment trend to plot, only a pipeline. This lens is structurally empty for a pre-revenue private; that emptiness is the finding.

---

Phase B — Measure performance

Lens 5 · Funding & Valuation Trajectory (+private swap for "Earnings Result")

The equity story is a round-trip down: peak hype → AV winter → down-round pivot.

• Total raised: ~$2.3–2.4B across A–E.
• The tell: valuation fell $8.6B → $6B (−30%) while the S&P and AI privates ripped — a hard down round, the market's verdict on the delivery-model failure.
• Cash / runway: Nuro held >$1B in the bank even at the 2022–23 trough and used two layoffs to stretch runway from ~1.5yr to ~3.5yr. Post-Series-E, management says the balance sheet funds "commercial expansion through 2027". Burn rate is undisclosed; the layoffs + repeated raises imply a historically heavy cash burn ($300M+/yr order-of-magnitude at peak headcount).
• Revenue: effectively immaterial. No licensing revenue figure is public; the Uber "close to $500M commitment" is an investment/purchase commitment routed via the deal, not Nuro P&L revenue. Treat current revenue as ~$0 at scale.

Read: a company that raised at the top, spent ~$1.5B proving the wrong business, and is now betting the down-round comeback on licensing. The 2027 runway is a hard clock (Lens 11/13).

Lens 6 · Founder & Management Voice (+private swap for "Earnings Calls")

No earnings calls exist. Substitute: founder/exec commentary in press + interviews.

• Consistent message post-pivot: "capital-light licensing," "AI-first / end-to-end," and the "classic second-mover" framing — Ferguson: "There is a lot of value in this sort of classic second-mover perspective… we get to learn from Waymo's stumbles and missteps". Tone: confident, technically credible, explicitly humble toward Waymo (a smart positioning move — no over-claiming).
• Shift in emphasis over time: 2021 = "free delivery for consumers, charge merchants" (a consumer-ops vision) → 2024–26 = "we are an autonomy supplier; Uber owns the fleet." The narrative pivot is coherent and, unusually for AV founders, de-hyped — they stopped promising imminent ubiquity and started promising a "very useful day-one" with a "broad operational design domain".
• Governance signal: Ferguson elevated president → co-CEO on 22 Oct 2025, explicitly to own capital-raising and commercial partnerships, while Zhu owns product/tech. Splitting the CEO role along "sell/raise" vs. "build" lines at exactly the moment the whole thesis becomes commercial execution is a rational, if telling, admission that fundraising and BD are now the gating function — not the tech.

What they've stopped saying: timelines for their own delivery robots; grocery/food TAM. What's recurring now: "second mover," "AI-first," "vehicle-agnostic," "through 2027."

Lens 7 · Cap Table & Secondary Marks (+private swap for "Comps")

Syndicate quality — genuinely tier-1, which is the strongest single bull data point. Crossover/mutual-fund presence is the IPO-proximity tell the overlay looks for, and Nuro has it in depth:

• Crossover / public-market funds: T. Rowe Price, Fidelity, Baillie Gifford, Tiger Global, XN — all present. This is the "IPO-ready cap table" signature.
• Strategics: SoftBank, Google, NVIDIA, Uber, Kroger, Woven/Toyota — an unusually strong strategic bench; NVIDIA + Uber joining the down-round Series E is a real vote (they have skin and commercial dependence).

Secondary marks — softly negative and drifting the wrong way:

• Forge implied share price ~$8.45 (2026-06-17); Nasdaq Private Market ~$8.64 (2026-05-22); ~$8.77 (Apr 2026). The trend across H1 2026 is down (~$8.77 → ~$8.45) — the private market is marking Nuro softer even as Waymo re-rated to $126B. That divergence is the market telling you Nuro is not being priced as a robotaxi winner.

The comp that matters most (peer-valuation reality check):

P/E, EV/Sales, ROE: n/a / not meaningful (pre-revenue private; no public multiples exist for Nuro). The honest comp is valuation-per-proof: Waymo commands $126B on 200M miles + real revenue; Nuro carries $6B on a pipeline. The WeRide/Pony.ai IPO haircuts (−22%/−47%) are the single most important comp for Nuro's own eventual liquidity mark.

Lens 8 · Catalysts That Moved the Mark (funding/product events, per +private)

No public stock; "moves" = valuation resets and deal events:

• Nov 2021 — Series D at $8.6B: the peak; AV/SPAC-era euphoria.
• Nov 2022 & May 2023 — two layoffs (20% then 30%, ~640 total; headcount 1,400 → ~700): the AV-winter reset; runway-extension mode. This is when the equity narrative broke.
• Sept 2024 — the licensing pivot: the strategic reset that saved the story.
• Apr 2025 — Series E at $6B (down round): the market re-priced the pivot 30% below peak.
• Jul 2025 — Uber/Lucid 20k-vehicle deal: the single most valuation-relevant event; converted "AI-first licensor" from slideware into a named contract.
• Aug 2025 — NVIDIA + Uber join Series E: strategic validation on the down round.
• Oct 2025 — Ferguson → co-CEO (commercial focus).
• Jan 2026 — CES production-intent Gravity robotaxi reveal + NVIDIA Thor; May 2026 — CA driverless testing permit.
• Apr 2026 — Uber/Lucid expanded to 35k vehicles + $500M Uber-Lucid.

Pattern: the mark reacts to (a) the funding environment and (b) named commercial deals — not to autonomy-tech milestones (Nuro has plenty of those and the valuation still fell). The market prices commercial proof and cash runway, which is exactly what Nuro most lacks.

---

Phase C — Judge people & books

Lens 9 · Management

• Track record: Zhu & Ferguson built core Waymo capabilities (perception, simulation, prediction) — elite, quantified by 100+ patents each and Ferguson's DARPA win. But at Nuro their operating record is mixed: they built a technically impressive delivery robot and a clean safety record, yet failed to build a viable delivery business — burned ~$1.5B, cut headcount in half, and had to abandon the original model. Credit for the self-aware pivot; debit for needing one.
• Skin in the game: founder-led, large founder stakes (each took ~$40M Google payouts into the venture; long tenure since 2016). Insider ownership specifics: n/a — private, not disclosed.
• Capital allocation: the weak spot. They raised at the top ($8.6B) and deployed it into a model they later scrapped; the Ike Robotics acquisition (2020 trucking) was quietly absorbed with little to show. On the other hand, they preserved >$1B of cash through the winter rather than blowing up — disciplined downside management. Net: poor strategic allocation, good liquidity defense.
• Red flags (governance): the dual co-CEO structure can diffuse accountability; splitting it at the commercialization inflection reads as "we need a full-time fundraiser/BD principal," a candid signal that capital access is the binding constraint. No related-party or comp scandals surfaced.
• Archetype: technical founders, not operators. Perfect for the R&D and licensing-IP phase; the open question is whether founder-engineers can run a commercial licensing business whose success now depends on partner management and sales, not code.

Lens 10 · Forensic Red Flags

Forensic lens on a private with no audited statements — so this is qualitative risk, not statement forensics.

• No audited financials at all. Revenue, margin, burn, cash — all founder/press-sourced and unaudited per public sources. That opacity is itself the headline forensic risk: outside investors are marking to management narrative + secondary trades, not to a P&L.
• Revenue quality: with ~$0 real revenue, there is no rev-rec risk yet — but the future licensing model has a classic recognition question (upfront license vs. per-mile vs. milestone) that will matter enormously the day it's booked. Watch for revenue that is really Uber/Lucid purchase commitments recycled as "bookings."
• Cash vs. narrative: the "runway through 2027" claim is unverifiable; two prior layoffs show the burn/runway math has already been wrong once (2025 runway was the 2022 promise, then re-cut). Treat "through 2027" as best-case, not covenant.
• Related-party / circular-financing flag: NVIDIA is simultaneously Nuro's compute supplier and a Series E investor; Uber is simultaneously the fleet operator, a customer's investor (Lucid), and a Nuro investor. These are legitimate strategic tie-ups but create circular value flows — an investor buying the product it funds. Not fraud; a valuation-quality caveat (the mark is partly set by counterparties with commercial motives).
• Dilution: five priced rounds + a down round = meaningful employee/early-investor dilution; option-holder value was reset by the $8.6B→$6B mark.

Regulatory findings (required sub-section). Per companies/nuro/regulatory/regulatory-findings.md (generated 2026-07-07): Nuro has no CIK — no SEC EDGAR search possible; total SEC findings = 0. Non-SEC web check run ("Nuro" (FTC OR DOJ OR FDA OR NHTSA OR settlement OR fine OR penalty) enforcement):

• No enforcement actions, consent decrees, or fines surfaced against Nuro.
• Safety: claimed 1.4–1.7M L4 miles, zero at-fault incidents; some minor non-fault low-speed contacts initiated by other road users, all resolved without injury and reported per state rules. NHTSA has an active review of AV exemption processes (sector-wide, up to 2,500 vehicles/mfr/yr rule), not a Nuro-specific investigation. TechCrunch (2026-03-31) notes Nuro — like all robotaxi firms — declines to disclose remote-assistance/disengagement frequency, a sector-wide transparency gap.
• Conclusion: No material regulatory or legal findings — verified via the SEC-EDGAR limitation note (private, no CIK), web enforcement search, and public safety disclosures as of 2026-07-07. Clean sheet, unaudited.

---

Phase D — Project & stress-test

Lens 11 · IPO-Readiness & Path-to-Tradeable (+private swap for "Forward Projection")

No EPS model — a pre-revenue private. The question is: when does this become a tradeable security, and at what mark?

IPO-readiness score: ~3 / 5 (late-stage, secondary-active; not pre-IPO/S-1 imminent) on the private-watch scale (1 early → 5 S-1 filed). Rationale:

• For (drivers toward an S-1): tier-1 crossover cap table already in place (T. Rowe/Fidelity/Baillie Gifford/Tiger) — these funds want liquidity; named marquee contract (Uber/Lucid); NVIDIA/Uber strategic halo; a clean safety/regulatory record. If SF robotaxi launches and a second licensee signs, an S-1 in 2027–2028 is credible.
• Against (what blocks the S-1): zero commercial revenue, one un-launched customer, runway only stated "through 2027," and a brutal comp set — WeRide/Pony.ai IPO'd at −22%/−47% to last private round; a Nuro IPO today would likely price at or below the $6B mark, which the crossover holders will resist. Public markets in 2026 pay for robotaxi revenue and miles (Waymo $126B) — Nuro has neither.

Milestones that unlock a tradeable event (the be-early watch-list):

1. CPUC driverless ride-hailing permit + DMV deployment permit (California) — the gating regulatory stack beyond the May-2026 testing permit.
2. Actual paid SF robotaxi launch (target 2026) — converts pipeline → revenue.
3. A second licensee (an OEM signing Nuro Driver Assist L2++, or a non-Uber robotaxi buyer) — de-risks single-customer concentration; the biggest re-rating catalyst.
4. Disclosed licensing revenue run-rate — the first real number.

Base / Bull / Bear on the mark (not EPS):

• Bear: SF launch slips into 2027, no second licensee, cash tightens → next raise is a flat-to-down round below $6B or a strategic acquisition (à la Zoox by Amazon). Plausible.
• Base: SF launches late-2026 at small scale, Uber/Lucid ramps slowly, one L2++ licensee signs → holds ~$6–8B into a 2027–28 IPO priced at a discount to private mark (WeRide/Pony.ai precedent).
• Bull: SF launch works, Uber scales toward the 35k commitment, 2–3 additional OEM licensees adopt the capital-light stack → re-rate toward $12–20B as the "arms dealer of L4" narrative takes hold vs. Waymo's vertical model.

No forecast.ts create in watchlist mode, and there is no EPS/binary to log — the trackable prediction would be "Nuro launches a paid, driverless (no-safety-driver) robotaxi service in San Francisco by 2026-12-31," which is the single falsifiable line to grade this dossier against.

Lens 12 · Bull vs Bear

Bull case. Nuro made the right pivot at the right time. Owning-and-operating AV fleets is a capital furnace that killed Cruise and is bleeding Zoox; licensing is the structurally superior model — Nuro monetizes the one thing it's world-class at (the autonomy stack) while Uber and Lucid absorb the capex. It arrives as a credible second mover with an AI-first, end-to-end architecture (NVIDIA Thor, foundation-model driving) that can generalize faster and cheaper than Waymo's older HD-map-heavy approach, and it carries a spotless safety record + regulatory-first status that buys OEM and permitting trust. The Uber/Lucid 35k-vehicle commitment is a genuine anchor few AV startups can claim, and the crossover-heavy cap table + NVIDIA/Uber strategic investment signals IPO-grade backing. If the capital-light thesis proves out with even 2–3 licensees, Nuro is the "Intel/Mobileye of robotaxis" and $6B looks cheap. Contrarian bull: the market is over-indexed on Waymo's lead and refuses to see that the winner of the AV race and the winner of the AV-supplier race are different companies — vertical integration may lose to a horizontal arms dealer once the tech commoditizes.

Bear case. Three ways it permanently impairs:

1. The customer never scales. The entire equity is one contract (Uber/Lucid) that has produced zero paid miles, on a car built by a going-concern-adjacent EV maker (Lucid), sold through an aggregator (Uber) that also backs Waymo/Wayve/WeRide/Pony.ai. If SF slips or Lucid stumbles, Nuro has no other revenue leg.
2. The licensing model may not clear at a price that matters. OEMs are notoriously brutal on supplier margins; a "license the brain" model competes with Mobileye (scaled, public, now going vertical to capture more value), Wayve (SoftBank-funded, same AI-first pitch), Tesla (free FSD data flywheel across millions of cars), and Chinese stacks. Nuro could win the deal and still earn a thin, delayed royalty.
3. Cash clock. Runway "through 2027" against a heavy historical burn and no revenue means Nuro likely needs another raise by 2027 into a market that just haircut WeRide/Pony.ai on IPO — a flat/down round or distressed sale is a live scenario.

Pre-mortem (18 months out, thesis broke): it's early 2028. SF launched thin in 2027, Uber prioritized its Waymo integration (which actually scales), no second licensee signed, Lucid diluted itself again, and Nuro raised a down round at ~$4–5B or sold to a strategic. The obituary line: "a great autonomy team that built a product nobody would pay enough for, fast enough."

Are multiples too high? There are no earnings multiples — but $6B on ~$0 revenue and one un-launched customer is priced for the bull case to substantially work. Downside to the mark is real (WeRide/Pony.ai/Cruise all say so).

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull:

• Revenue concentration is ~100% in a deal that hasn't started. Name one dollar of scaled licensing revenue. You can't. The bull case is a pipeline wearing a valuation.
• The moat is a safety slide, not a defensible asset. 1.7M miles is a rounding error next to Waymo's 200M autonomous miles and 500k paid rides/week — Waymo's data + operational flywheel compounds daily while Nuro's sits idle pre-launch. "Zero at-fault incidents" over a tiny mile base is survivorship, not proof of superiority.
• The most dangerous competitor bulls underestimate: Mobileye — and Tesla. Mobileye already ships driver-assist at automotive scale, has OEM relationships Nuro can only dream of, is public with real revenue, and is now going vertical to capture more of the stack — squeezing horizontal licensors from above. Tesla arms itself with free fleet data. Nuro is a sub-scale licensor caught between a scaled supplier and a data monopolist.
• Capital-allocation history is poor. Management raised $8.6B and spent ~$1.5B proving the wrong model; the Ike acquisition vanished. Why trust the next $2.4B thesis?
• Circular financing flatters the mark. NVIDIA (supplier) and Uber (operator/customer's backer) investing in the down round is commercially motivated validation, not disinterested price discovery. Strip that and the secondary market is marking Nuro down through 2026.
• What must hold for $6B: SF launches paid & driverless on time (2026), Lucid ships Gravity at quality, Uber prioritizes Nuro over Waymo, and a second licensee signs. Four contingent "ands."
• If growth disappoints 20–30% (i.e., launch slips a year / one licensee only): the next raise is flat-to-down; option value evaporates; a $4–5B strategic sale (Zoox precedent) becomes the base case.
• The single scenario that permanently impairs: Lucid financial distress or an Uber pivot fully to Waymo — either kills the only commercial leg overnight, through no fault of Nuro's technology. Plausibility: moderate-and-rising, given Lucid's own balance sheet and Uber's multi-partner promiscuity.

Lens 14 · Fifteen Questions for Management (ordered by information value)

1. What is the actual licensing revenue model — upfront license, per-vehicle, per-mile royalty, or milestone — and what is the expected lifetime revenue per Nuro-Driver-equipped vehicle? (This single answer determines whether the business is a $50M or a $5B revenue line.)
2. Beyond Uber/Lucid, how many signed, revenue-bearing licensees do you have today, and how many by end-2026? (Concentration is the whole risk.)
3. What is your current monthly cash burn, cash balance, and the precise month "runway through 2027" runs out — and what raise/terms do you assume beyond it?
4. If Lucid cannot ship the Gravity at target volume/quality, what is your fallback vehicle — and how long to re-integrate the Nuro Driver on a different OEM's line?
5. Uber partners with Waymo, WeRide, Pony.ai and Wayve. What contractual volume commitment or exclusivity protects Nuro from being deprioritized as Uber scales Waymo?
6. What is your remote-assistance / disengagement rate today — the metric every robotaxi firm, including you, currently refuses to publish?
7. How does your unit economics for OEMs beat Mobileye's, given Mobileye's scale, and how do you respond to Mobileye going vertically integrated?
8. Precisely which permits (CPUC driverless ride-hail, DMV deployment) remain before a paid, no-safety-driver SF launch, and what is the realistic date — not the target date?
9. What is your defensible moat against Tesla's fleet-data advantage as end-to-end/foundation-model driving commoditizes the approach you're selling?
10. You wrote off ~$1.5B and half your headcount on the delivery model. What specifically did you learn that makes the licensing thesis different, and what would make you kill this model too?
11. NVIDIA is both your sole compute supplier and a shareholder, and arms your competitors on the identical Thor platform. Where is your differentiation if the hardware is common?
12. Why the dual co-CEO structure at the commercialization inflection, and how are strategic accountability and capital-allocation decisions actually split?
13. What is the L2++ Driver-Assist go-to-market — any signed OEM — and does it realistically bridge cash before L4 robotaxi revenue scales?
14. What return on the ~$2.4B invested do you underwrite for a 2027–28 IPO, given WeRide and Pony.ai priced at 22–47% discounts to their last private rounds?
15. In your own pre-mortem, what is the single most likely reason Nuro is worth less in 2028 than the $6B mark today?

---