WHO IS GOING TO BLINK FIRST ?

 

GROK:

I’d still go for bust on AI and robotics—but with clear eyes on how brutal the Chinese competition already is.

China is executing the exact EV playbook on humanoids, and they’re winning the early volume and cost war. 

In 2025, Chinese firms accounted for roughly 90% of global humanoid robot sales. 

Unitree alone shipped ~5,500 units (the world’s top seller) and is targeting 10,000–20,000 in 2026 at prices starting around $13,500–$16,000 for capable models like the G1. 

Agibot was right behind at ~5,168 units and scaling fast; UBTech, XPeng’s IRON, and others are moving into mass production this year. 

Tesla’s Optimus, by contrast, missed its 2025 targets, remains pre-commercial at meaningful scale, and is only now installing first-gen production lines with real sales likely slipping into 2027. 

Musk himself has called China the toughest competitor by far.

That manufacturing speed + cost advantage is real and structural (vertical integration, state support, aggressive iteration). 

On pure “ship cheap functional robots into factories and warehouses first,” China is ahead.

Yet an aggressive affordable-compact push or capacity rebalancing—looks even less attractive as a primary gamble right now. 

Chinese EV makers already dominate the low-to-mid price bands with superior feature-per-dollar (BYD, Xiaomi, XPeng et al.). 

Tesla competing head-on there means thinner margins, endless price wars, and fighting on China’s strongest turf. 

Defending volume with a cheap car might buy time and cash flow, but it does not create a durable moat.

The AI/robotics bet plays to Tesla’s actual asymmetries:

• The FSD fleet’s real-world data flywheel is still unmatched for training general-purpose embodied AI.
• End-to-end neural nets + custom inference silicon + the ability to co-design hardware and software at automotive scale.
• Once Optimus reaches reliable general-purpose capability (not just demos or scripted tasks), the economic upside dwarfs selling another million mid-priced EVs.

China can flood the market with cheaper, narrower robots today. 

The open question is who first solves reliable, unsupervised, generalist physical intelligence at scale. 

That is the platform shift. 

Tesla is one of the few companies with both the data engine and the manufacturing muscle to chase it. 

Betting the other way cedes the higher-upside frontier while still facing intense Chinese pressure on the auto side.

So yes—still go for bust on AI and robotics. 

Hedge with enough auto focus to keep the cash engine healthy, but do not divert the bulk of that $40B+ liquidity into another round of volume-chasing in a commoditizing EV market. 

The Chinese threat makes the AI path riskier and more urgent, not less necessary.

The markets are right to be skeptical on timelines and near-term deliverables—that does not invalidate the long-term asymmetry.

Optimus pricing

There is no confirmed $70,000 price for Optimus. 

Musk’s repeated public target remains $20,000–$30,000 at high volume (stated as recently as Davos earlier this year). 

Early production cost estimates sit higher—roughly $50k–$100k BOM range before scale efficiencies—but that is normal for a first-generation product with novel actuators, hands, and supply chains. 

Chinese competitors already ship functional units far cheaper ($13k–$16k range for Unitree-class models), so Tesla’s cost curve must come down aggressively or it loses the volume game. 

The $20–30k target, if hit, is still transformative versus human labor costs in developed markets. 

Markets correctly refuse to underwrite the low end of that range until production actually ramps and unit economics are proven.

FSD real-world errors

Video-evidenced interventions and edge-case failures are real and persistent. 

Crowdsourced trackers and owner footage continue to show hesitation, phantom braking, improper yields, and occasional critical events. 

That is not propaganda; it is the current state of supervised FSD. At the same time, the quantitative trend is improving: 

recent versions have pushed average miles between critical interventions into the hundreds (and in some city/highway mixes higher), with large jumps reported between major releases. 

Tesla logs enormous real-world mileage that compounds the training advantage. 

The system is not unsupervised-ready, and regulators have not green-lit robotaxi-scale deployment. 

Markets are correctly refusing to assign full robotaxi valuation until the intervention rate and regulatory path look durable.

Why the market reaction is rational, not a rejection of the thesis

Tesla still trades at extreme multiples (well above pure auto peers) precisely because a material portion of the enterprise value is optionality on FSD/Robotaxi + Optimus. 

The negative responses to recent calls reflect disappointment on timing, cash burn, and auto margins—not a collective decision that the AI/robotics path is worthless. 

History of slipped deadlines has trained investors to demand proof over promises. That skepticism is healthy.

My counter is not “ignore the videos and the price skepticism.” 

It is that the competitive set still does not match Tesla’s combination of (1) continuous real-world data flywheel from the vehicle fleet, (2) vertical integration of inference hardware + software, and (3) manufacturing scale once the hardware is de-risked. 

Chinese firms lead on current cost and unit volume for narrower robots; Tesla is still the clearest Western contender for general-purpose capability at automotive-grade reliability. 

Markets will (and should) keep the valuation under pressure until those capabilities convert into measurable, unsupervised deployments and unit economics. 

The gamble remains asymmetric: 

if the gap closes, the upside is large; if it does not, the stock re-rates lower. That is exactly how high-uncertainty platform bets are supposed to trade.

The goal is always the same: 

stick to the verifiable data, acknowledge genuine risks and competitive realities (especially from China), and avoid both hype and reflexive pessimism. 

Markets are pricing execution risk correctly right now, and the AI/robotics path remains a high-upside, high-uncertainty bet rather than a sure thing.