Tesla Is Making a Big Tech-Sized AI Bet Without Big Tech Margins
PM Wong’s National Day Rally speech on August 23 is what got me looking at Tesla (NASDAQ: TSLA) again.
Two-thirds of Singapore’s taxi drivers are over 60, and bus captains are becoming harder to recruit. What struck me was that his case for autonomous vehicles was really a labour argument rather than a technology one. That made me more confident in the robotaxi theme generally, since I already hold a few names exposed to it.
Tesla is probably the most obvious US-listed way to play the same theme. So I decided to look at whether I would actually buy the stock today, rather than simply getting excited about another Cybercab milestone.
The labour issue also feels quite real to me living in Singapore. Not many younger Singaporeans seem interested in driving taxis or buses for a living any more, and you can see the manpower pressure on the ground, with operators increasingly relying on drivers from Malaysia, China and elsewhere. AVs are not really solving some distant future problem here. They are addressing a manpower issue that already exists.
It’s probably worth explaining why this hits differently here than it would in the US. Most Americans drive their own car, so a taxi or bus driver shortage is somebody else’s problem. In Singapore, owning a car isn’t really an option for most households — the Certificate of Entitlement alone, just the permit to register a vehicle for ten years, hit a record S$129,000 in July 2026, on top of the car itself. A mass-market sedan like a Toyota Corolla, which would run about US$22,000 in America, now comes out closer to S$200,000 — roughly US$155,000 — once COE, registration fees and taxes are all added in. So public and shared transport isn’t a lifestyle choice for most of us, it’s the only realistic option, which is exactly why a driver shortage becomes a national policy issue rather than a personal inconvenience.
I don’t own Tesla. After going through the numbers, I’m still not buying it, although not for quite the reason I expected.
The numbers first
Tesla delivered 480,126 vehicles in Q2 2026, up about 25% year-on-year. After a period of weaker deliveries, that was a good quarter.
The rest of the numbers were less encouraging.
Automotive gross margin excluding regulatory credits fell from 19.2% in Q1 to 16.3% in Q2. Company-wide operating margin came in at just 1.4%, down from 5.7% in Q4 2025, leaving Tesla with only about $0.4 billion of GAAP operating income for the quarter.
GAAP net income was $1.11 billion, or $0.32 a share, but that number included a $1.0 billion unrealised gain on Tesla’s SpaceX stake. For me, the operating income figure gives a cleaner picture of how the underlying business performed.
Free cash flow was negative $1.1 billion for the quarter, while capex jumped to $5.8 billion as Tesla continued investing in AI compute, its own chip fabrication plans and Cybercab production.
None of this suggests Tesla is in financial trouble. It still has more than $43 billion in cash.
What it does show is that the business which currently earns most of the money is becoming less profitable at the same time Tesla is spending more aggressively on businesses that are not contributing much profit yet.
Same playbook as Big Tech, thinner margin underneath
In some ways, Tesla is doing something similar to Meta, Google, Microsoft and Amazon.
They are all spending heavily today on AI infrastructure and businesses they expect to become much larger later. Investors are being asked to accept higher spending now in exchange for potentially much larger profits in the future.
The difference, to me, is where that spending is coming from.
Meta’s operating margin is still above 40%. Even after spending heavily on AI, there is a very profitable core business underneath it.
Tesla is starting from a much thinner base. Its company-wide operating margin was only 1.4% last quarter, while automotive gross margin excluding credits fell to 16.3%.
That does not mean Tesla cannot afford the investment. It clearly can, especially with the cash it has on the balance sheet.
But it does mean the margin for error is smaller.
If robotaxi, Optimus or Tesla’s AI infrastructure take longer than expected to produce meaningful returns, there is much less profitability from the existing business to absorb that delay.
That is what makes the current investment phase more uncomfortable for me than the headline capex number itself.
The business that still pays the bills is also changing
Tesla announced in January 2026 that it would end Model S and Model X production, with the final cars rolling off the Fremont line in May. That space is now being repurposed for Optimus production.
Management’s recent earnings commentary also spends far more time discussing Optimus, Cybercab and AI compute than the traditional vehicle business. Musk has described Optimus as potentially the company’s biggest product.
I don’t necessarily think that is the wrong long-term decision.
If Tesla really believes robotics and autonomy are much larger opportunities than selling cars, it makes sense to allocate capital accordingly.
But as an investor, I still have to ask what funds that transition in the meantime.
Today, the car business is still doing most of the heavy lifting. Yet its margins are already coming down while more capital and management attention move elsewhere.
The market seems to assume Tesla can move from EV profits to robotaxi and robotics profits without too much disruption along the way.
I am less certain that the transition will be that smooth.
What the share price is telling me
TSLA closed at $348.75 on 28 August, down roughly 21% year-to-date and about 30% below its all-time high of $489.88 set in December 2025.
The 52-week range is roughly $297 to $499, which puts the current price much closer to the lower end of that range than the top.
The stock fell after the Q2 earnings miss in July, but it did not continue falling aggressively afterwards. It has mostly traded within the $300-$360 range and still reacts strongly to macro news, particularly interest-rate expectations.
That tells me Tesla is still being valued mainly as a long-duration growth company rather than simply as an automaker with weakening margins.
Average analyst price targets sit around $390, although the range is enormous, from roughly $125 to $600.
I don’t find the average especially useful when the assumptions behind those estimates are so different.
Tesla is one of those companies where getting the autonomy assumption wrong can overwhelm almost everything else in the valuation. I suspect that explains at least part of the very wide range of targets.
It is also worth putting an actual number on what “not cheap” means.
At $348.75 as of 28 August, Tesla is worth roughly $1.38 trillion. That is around 13 times trailing revenue, more than 360 times trailing earnings and about 183 times forward earnings.
Those multiples do not tell us what robotaxi or Optimus will eventually be worth. Today’s earnings hardly come from those businesses.
But they do tell me one thing: I would already be paying quite a lot today for the possibility that those businesses become very valuable later.
That is where I struggle with the stock.
The current price does not look obviously cheap when margins are still falling. At the same time, I would not call it obviously expensive either when the stock is already well below its highs and Tesla’s progress in autonomy is real.
That leaves me somewhere in the middle.
I don’t see enough margin of safety to buy it, but neither do I see enough evidence that the market is badly overestimating Tesla’s future to bet against it.
That contradiction isn’t unique to Tesla, either. The China-listed robotaxi names I hold have fallen more than 30% this year despite being further along on monetisation than Tesla is. Some are already disclosing fare revenue, fleet economics and even unit-economics breakeven in individual cities — information Tesla still does not provide. The market hasn’t rewarded that disclosure with a better share price. So I don’t think “further along on monetisation” is enough on its own to protect a stock in this sector right now, which is a useful check against assuming Tesla’s price problem is somehow a Tesla-specific problem rather than a sector-wide one.
What would change my mind
The first thing I would watch is automotive gross margin excluding credits.
I would like to see it stabilise above 17-18% for a couple of quarters. That would give me more confidence that the existing business has found some kind of floor while Tesla continues spending heavily elsewhere.
The second thing is robotaxi scale.
This is one area where I have already had to update my own view.
Earlier in the summer, Tesla’s Austin fleet was still only around 20-30 active vehicles. By late August, crowdsourced tracking from Robotaxi Tracker, cited by The Verge, had identified 54 different Tesla vehicles completing 170 unsupervised rides over two weeks. Every tracked ride in that period reportedly had no safety monitor on board.
That is genuine progress.
Tesla has also expanded its footprint across seven metros, including Austin, Dallas, Houston, Miami, Orlando and Tampa, together with a Bay Area service that still requires a safety driver.
So I probably need to be more balanced here than I was initially. Tesla is scaling faster than I first thought.
The problem is that scale is still relative.
Tesla has not disclosed enough information on ride volumes, utilisation or cost per mile for investors to understand the economics of the service. Musk also said on the Q1 2026 call that robotaxi would not be a meaningful revenue contributor this year.
For me, that is the more important issue.
I don’t really need Tesla to show me that it can put more cars on the road. I need enough information to understand whether each additional car is moving the business closer to sustainable economics.
Waymo is useful as a comparison because it is already operating at a much larger commercial scale.
I’ve ridden one myself in San Francisco and wrote about the experience previously when I covered WeRide.
What I remember most was how smooth the ride was going up San Francisco’s hills compared with the Uber rides I took during the rest of the trip. Obviously that is just one person’s experience, but it helped me understand why people might actually use the service repeatedly rather than treating it as a novelty.
The operating numbers are much larger too.
Waymo runs more than 700 vehicles in San Francisco alone, with additional fleets in Los Angeles and Phoenix. It operates fully driverless services across 11 US metros and is doing roughly 500,000 paid rides a week.
Amazon’s Zoox sits somewhere in between.
It only started charging fares in Las Vegas this August, its first commercial market, while San Francisco, Austin and Miami are still offering free rides. Its fleet is around 100 vehicles across Las Vegas and San Francisco, it has logged about 3 million miles and roughly 1 million riders cumulatively, and it is building a Hayward factory designed to produce up to 10,000 vehicles a year by 2027.
Tesla is therefore moving, but it is still some distance behind operators that have already accumulated much more operating history.
The comparison also reminds me that robotaxi progress should not be judged only by how many cities a company enters. Deployment can move quickly. The unit economics usually take much longer.
On the downside, if Tesla’s automotive margins continue falling while capex keeps rising, I would become more cautious rather than simply waiting for a better entry point.
I don’t think this gets resolved quickly
Every robotaxi company I follow, including Tesla, is probably still a few years away from proving that the business can be consistently profitable at fleet level.
That is consistent with what I have written about this theme before.
The deployment numbers often move much faster than the economics.
So buying Tesla today, or any robotaxi stock today, means paying for a business that may not properly prove itself for several years.
You have to be comfortable holding through many quarters where the spending shows up before the profits do.
I’m not against making that kind of investment. In fact, I have already done it elsewhere within the autonomous-driving theme.
I just don’t think Tesla’s current valuation gives me enough room to take the same bet here.
Bottom line
Singapore’s National Day Rally reinforced why I like the robotaxi theme. It still did not make me want to buy Tesla.
Tesla’s valuation has felt expensive to me for years, even before I looked closely at this quarter. What the latest numbers did was give me a clearer reason why I remain uncomfortable with it.
The car business is still funding much of the journey, but its margins are getting thinner while Tesla spends more on robotaxi, Optimus and AI infrastructure.
Those businesses could eventually become much bigger than the car business. I am not ruling that out.
But at a valuation of roughly $1.38 trillion, I would already be paying quite a lot today for that future.
Maybe Tesla gets there. I would not bet against it.
There’s also a portfolio-level reason I’m not rushing into this. I’ve already watched the China-listed names I own in this theme fall more than 30% this year despite their monetisation progress. Adding Tesla now would mean adding another volatile robotaxi position whose economics may still take years to prove. Given the exposure I already have, I don’t see much reason to rush into another version of the same bet.
For now, I’m watching from the sidelines.
This is not financial advice. The author may hold positions in securities discussed. Readers should conduct their own due diligence and consult with a qualified financial advisor before making investment decisions.