grep

Anything to distract me from life. Professional Cred: Tech Exec@ @amazon, Edu: MBA @ChicagoBooth, Mich Compsci, Interests: AI/ML, EVs, Astronomy

[grep @avarshne]

This is one of the bigger accomplishments of 2026 then most people realize. Congrats @SpaceX & @elonmusk !

[Sawyer Merritt @SawyerMerritt]

SpaceX's Starship rocket has successfully reached orbit and deployed its V3 Starlink satellites into orbit for the first time, officially making this Starship's first revenue-generating flight, a major milestone for the company!

Congrats @SpaceX team! Incredible achievement 🚀

[grep @avarshne]

This is all experimental pricing by Tesla so all of this is moot! One should compare when there is enough supply of driverless RoboTaxis in service in the Austin area. At steady state- Dynamic pricing is comparable. If the promised unit economics of CyberCab hold, the price should be within 40-60% of Uber’s.

[grep @avarshne]

What is other option? Ship milk from India over 30 days on a container or on an expensive plan route? BTW, Coke/Pepsi/Unilever they all do local sourcing/global branding. This is quite common.
The target market for Amul is the Indian diaspora which is buys disproportionally at Costco.

[grep @avarshne]

This is a bad analogy. Meta, xAI, Google and some of the frontier labs have all the ingredients to build a "proactive" agent like Muse Spark so any one can implement. These things change every few weeks - remember Open Claw hype few months ago. Muse Spark is great but let's check on it few weeks from now.

Now the punchline - That is not the case for Autonomy. All the players except for Google/Waymo don't have the same ingredients. Apologies for being condescending - stop asking "industry experts" and think from 1st principles yourself before you make these statements or stop copy pasting from Bloomberg terminal.

[grep @avarshne]

Once unsupervised FSD becomes available, the use cases and productivity possible within your Tesla goes exponential. Voice Assistant + Screen give you all kinds of capabilities (1) Shopping (2) Interactive prep/study (3) Video conferencing (4) Entertainment (5) Sleeping :-) and list goes on..

The 1hr in commute suddenly becomes your most productive hour.
@wholemars

[grep @avarshne]

Cybercab is a utility vehicle - get you from pt A => pt B. I doubt there will be any ownership model except for Fleet operators.
Here are my 3 predictions of how autonomy will become mainstream in US/Canada (timeframe 2-3 years).
1. Per ride payments similar to Uber/Lyft- Predominant model.
2. Monthly subscriptions which give you certain number of rides/week e.g. $100 for 10 rides within 10 miles geo-fences. Appeals to commuters.
3. Weekend Rental - a Model Y/3/CyTruck with a steering wheel and FSD Supervised (potentially unsupervised in the future) will come to you. Take it outside the geofences e.g. Tahoe, Yosemite or to your weekend getaway. Rates would be competitive to rental car companies since there is no cost of running a office and the customer experience will be way better with Pick & Drop off at your convenient place with FSD streamlining you drive.

[grep @avarshne]

$TSLA could be easily selling 700K - 800K deliveries / Qtr had they continued to invest in new models. This is all the while investing in RoboTaxi & Optimus. There is no opportunity cost here and anyone who tells you this is BSing and has never run a large company before.

FSD/RoboTaxis will take > 5/8 years to get to countries like India, China, SE Asia, so you need new Tesla models. Elon has a habit of taking un-necessary risks and making it look like he had no choice. Case in point - He wanted to kill the Falcon 9 program before Starship took off. Gwen Shotwell stopped him. (Quote from his Biography).

[grep @avarshne]

@garyblack00 @garyblack00 Can you share FSD awareness data across the US? Survey/Panel?

Assumption is also not causation !

[grep @avarshne]

@ankurnagpal Good selection of companies. Happy to have a small investment through USVC in these.

[grep @avarshne]

@wholemars Exactly ! Dynamic pricing is the best market mechanism to manage supply and demand. Tesla can unlock more CyberCabs to the roads as demand surges which in turn drives down prices.

BTW, at current scale, Tesla pricing is just experimental.

[grep @avarshne]

@npew Most of this makes sense. Why care about <$30K car and why not add better than P50 of human driving fleet as a criteria?

[grep @avarshne]

I call this the "Business OS" of India—and right now, it’s actively stopping India from building true global product companies. The same human capital when it comes to Silicon valley where I live thrives and you seem them exhibiting high risk taking behavior.

A country's Business OS is a mix of its culture, incentives, capital, and regulations. Ours basically punishes risk-taking.

- Think about it. Most Indian businesses operate in a protected bubble. They’re either govt-adjacent, local copycats, or just tech wrappers on service businesses (like food delivery).

- Building a global product requires massive risk, deep R&D, and original brand building. But there is zero incentive to do that here because our cost of capital is insanely high (>15%).

- This is exactly why the giants—Ambani, Adani, and worst of the lot, Birla—only invest where there’s govt assistance and guaranteed demand (telecom, ports, petrochem). If you’re guaranteed a 20% ROIC and only you can afford the 15% cost of capital, why bother taking real innovation risks?

- Here is the acid test for a country: the Top 20 companies by market cap should turn over every 10 years. In the US, the list changes almost every decade. That’s what a vibrant capitalist market looks like.

- Look at TCS, Infosys, or Cognizant. They’ve had bottomless capital and decent talent (mostly mediocre) for 40 years, yet haven't built a single globally renowned product. Why? The Business OS.

- Even the consumer "brands" are largely an illusion. It’s mostly 80% white-labeled product from China, repackaged, and slapped with a celebrity endorsement. You aren't building a global brand that way.

India need policies that rewrite this OS from the ground up:

1/ Drastically lower interest rates (2-3%) so the cost of risk capital isn't fatal. 2/ Gut the red tape. I should be able to register, clear all paperwork, and start almost any business within 30 days. 3/ Shift the incentives from services to R&D and product building. Products can be exported globally; services (outside of tech) cannot. Until we fix this, our export growth will stay stuck.

[grep @avarshne]

@Vigilante_1983 @OverlyTrev @elonmusk @Tesla_AI Significantly better AI than today will be needed. Not “AGI” as some people are purporting.

[grep @avarshne]

You are not wrong on the 5-7 year time horizon for Level 5 although you don't need AGI to do Level 5. A lot of the pieces you are referring to - energy efficient edge inference, low-cost chips, multi-modal driving specific models already exist today. They will get you to L4/4.5. Tesla, Waymo, Huawei are all getting close to this.

[grep @avarshne]

@bindureddy What is the implication for Anthropic & Open AI? At what point the API business for the Frontier labs start getting cannibalized.

Their enterprise subscriptions and associated token usage is driving their revenue but sooner, OSS will creep through this as well.

[grep @avarshne]

That is an incredible growth number. Personally, I believe this is an extrapolated number. For comparison, Uber does around 16M rides/mo in California so if Waymo is already at 9% that is a strong signal.

Unit economics at scale should be at steady state at this scale - rumors are close to $1.6 - $1.98 per mile. Now it is Tesla's turn to show what they can do.

[grep @avarshne]

@KirkNeet @garyblack00 Can a consumer go and purchase a RT6? That's the litmus test for the real price/cost. Getting autonomy to work in a PAV (Personal AV) is a significant challenge.

[grep @avarshne]

@FStirnkorb @itskyleconner Show me a Chinese car that has close to FSD like capability that you can purchase. Not talking about Robotaxis and don't go for demo videos. Xpeng is the closest. If I were to bet, Chinese will get there but will still take them 2-3 years.

[grep @avarshne]

I expect some push back on the range argument. For the non-fans segment, surveys show that range anxiety + charging are the top 2 concerns against EVs. I live in "Teslafornia" and even here, I have bunch of friends who exhibit this - looking to go for Hybrids from Toyota & Honda.

Back in 2021, Tesla promised between 3-5% / yr in battery density improvement - 5 year compounding should be close 20% density improvement. But, I don't see any of it showing up in the EPA ranges.

All Premium Model Y's should by have 420+ while CyTrucks close to 500 miles by now. The charging curve is another story. The argument around you need to stop every 2 hours anyways is stupid. A long range vehicle is not just for road trips but for folks who can only do public charging. For CyTruck towing or load bearing eats up range. Colder climates can decrease range by 40%. Tesla has justified its own engineering trade-off as enough for customer need - I call BS on that. They stopped innovating on battery and got stuck in the 4680 rut.

[grep @avarshne]

$TSLA has lagged the broader market (SPY) and not just the Tech index (QQQ). There has been no growth driver as is evident in its Cash Flow from Operations (not FCF). IMO, the key mistake is un-necessarily de-emphasizing Auto, Battery tech. Tesla would have been selling 2M+ vehicles globally at 20% EBIT margins with 400 miles+ range.

Elon in his typical style takes unnecessary risks by completely starving his vehicle line up for the sake of FSD, RoboTaxi and Optimus. I am all for these huge bets but he offers a false choice. The have enough capital (Human + Financial) to do multiple priorities at the same time. A $1T+ company ought be able to do multiple things at the same time.