Hey, it's April 23, 2026. — Episode 445

Hey, it's April 23, 2026.

Tesla just posted its Q1 numbers and the profit picture is a lot stronger than many expected, giving them real runway to push harder into AI, robotics, and the next wave of autonomy. The stock reaction was muted though, as the big capital spending plans and a tiny revenue miss left some traders wanting more clarity on the returns. It's a classic Tesla moment where the financials show progress but the market is already pricing in what comes next.

The spending side is getting a lot of attention today. Tesla is raising its 2026 capex outlook, with a clear focus on AI infrastructure and robotics projects that Elon sees as central to the company's future value. From a business standpoint this tells you they're choosing to invest through the current margin recovery rather than coasting. For the industry it reinforces how Tesla is doubling down on proprietary technology instead of playing defence.

On the product side, the 2026.2.9.8 update with FSD 14.3.2 is now officially rolling out. The release notes lay out the latest refinements in how the system handles complex driving scenarios. For owners this is the tangible side of all that AI investment — steady, incremental gains that should build confidence over time. It also shows how Tesla can improve vehicles already on the road without waiting for new hardware.

Tesla's European sales jumped sharply in March, up over 100 percent year-over-year after a string of softer months. That rebound comes despite ongoing competition from BYD and some political headwinds in certain markets. For the business it proves demand is still there when the product and pricing line up, and it gives breathing room while they sort out longer-term growth in the region.

Efforts are underway to bring Full Self-Driving to China as quickly as regulators will allow. The company said on the earnings call that the work is active and a priority. This matters because China represents both a massive addressable market for supervised autonomy and a critical testing ground where local data could accelerate improvement cycles.

The Semi is running into a more crowded field than it faced a couple years ago. Traditional truck makers and new EV entrants are bringing their own heavy-duty electric offerings, putting pressure on Tesla's timeline and feature set. It’s a reminder that while the Semi was an early bet, the commercial vehicle space moves fast once competitors commit real resources.

Tesla recorded a $173 million accounting loss on its crypto holdings in the quarter, though Bitcoin holdings themselves stayed unchanged. It’s a small line item in the bigger picture but highlights how external asset volatility can still swing quarterly results. For customers and investors it’s mostly noise compared to the operating business, yet it shows the balance sheet still carries some non-core exposure.

One interesting note from the earnings was the sharp rise in Full Self-Driving supervised subscriptions. That recurring revenue stream is becoming more material and gives Tesla a direct financial signal on how many owners trust the system enough to pay monthly. It’s one of the cleaner proof points that the software side is starting to scale.

During the call Elon highlighted progress on the Terafab manufacturing project, describing it as a key enabler for cheaper future vehicles and higher volume. The stock stayed relatively flat on the news, suggesting investors want to see concrete output before they fully price it in. Still, if Terafab delivers on cost reduction it could be a bigger deal for long-term margins than anything else discussed.

Volvo’s recent momentum in certain EV segments has some analysts asking whether traditional luxury brands are pulling ahead in execution where Tesla once led. It’s not an existential threat, but it does illustrate how the competitive bar keeps rising. Tesla’s answer will likely be a combination of software differentiation and the cost advantage they hope to unlock with the next generation of vehicles.

Short Spot

Tesla’s Electric Truck Faces Competitive Pressure: 23 April, 2026, 12:22 AM PDT, Intellectia AI

The Semi no longer enjoys a clear first-mover advantage. Rivals have narrowed the gap on range, payload, and charging infrastructure, forcing Tesla to prove the economics in real fleet operations rather than on paper. The challenge is honest: commercial buyers care about total cost of ownership more than brand hype. Tesla’s path forward is still strong if they can leverage their own data and energy ecosystem to create a lower operating cost that competitors can’t match. Source: news.google.com

Tesla First Principles

🧠 Tesla First Principles - Cutting Through the Noise

TOPIC SELECTION: Taking a step back from today's headlines, let's apply first principles thinking to what actually determines whether robotaxi networks become profitable at scale versus staying an expensive science project.

The Surprising Truth: Most debate focuses on regulatory approval or hardware capability, yet the physics and economics show the biggest variable is simply how many high-utilization miles a vehicle can deliver in a given geography before the energy and maintenance curves turn negative.

The Fundamental Question: At what combination of utilization rate, energy cost, and cleaning/downtime does a robotaxi fleet generate genuine cash flow instead of burning capital?

The Data Says: Real-world fleet data (even from human-driven ride-share) shows that once a vehicle sits idle more than about 60 percent of the day the economics collapse under depreciation and opportunity cost. Tesla’s advantage is vertical integration across energy, software, and manufacturing — they can attack each of those variables simultaneously rather than hoping one breakthrough fixes everything.

The Tesla Approach: Strip the problem to atoms: build the lowest possible cost per occupied mile by owning the energy supply, using software to maximize dispatch efficiency, and designing vehicles that need almost no human intervention between trips. Everything else — flashy demos, regulatory lobbying, brand marketing — is secondary.

The Bottom Line: If Tesla can keep pushing utilization upward while driving hardware and energy costs down, the network becomes profitable long before perfect Level 5 autonomy is required everywhere. That’s the part the conventional wisdom consistently gets wrong — it’s an optimization problem, not a binary technology switch.

That’s what stood out to me today. The earnings show Tesla has the financial oxygen to keep swinging big, but the real test is still execution on the fundamentals that actually move the cost curve. Drop me a note at @teslashortstime if anything here landed differently for you — always good to hear what you’re seeing.

Battery Chemistry Evolution – A Global Perspective

Let’s talk about how battery chemistry has quietly become one of the most important strategic choices an EV maker makes. Early lithium-ion cells were mostly variations on cobalt-based formulas that delivered decent energy density but came with high cost and thermal management headaches. As the industry scaled, two main paths emerged.

In Asia, particularly China and parts of Korea, manufacturers leaned heavily into lithium iron phosphate (LFP) chemistry. It gave up some range on paper but won on cost, safety, and cycle life. LFP packs could be produced with fewer critical minerals and ran cooler, which simplified vehicle design and improved longevity for fleet operators who rack up high mileage. This approach helped bring sticker prices down faster in price-sensitive markets and reduced reliance on materials that are geographically concentrated.

Western and some Japanese makers initially favoured nickel-manganese-cobalt (NMC) blends, chasing every possible kilometre of range to overcome consumer range anxiety in colder climates or regions with less charging infrastructure. NMC offered higher energy density, which translated into lighter vehicles or longer driving distance for the same weight. The trade-off was higher cost, more thermal engineering, and greater exposure to cobalt and nickel price swings.

Tesla has tried to thread the needle with its 4680 cell format. By moving to a larger cylindrical design and incorporating silicon in the anode, they’re aiming for a better balance of energy density, power, and cost. The structural pack approach also lets the cells themselves contribute to vehicle rigidity, reducing the need for heavy additional reinforcement. Internationally this is watched closely — European makers are studying how to adapt similar tabless designs for their own platforms, while Chinese producers are scaling LFP even further with sodium-ion experiments for entry-level cars.

The real-world performance differences show up in unexpected places. Fleets using LFP often report lower degradation after three or four years of heavy use, which matters when residual values determine financing rates. NMC vehicles sometimes deliver stronger cold-weather performance out of the box but can require more active cooling in hot climates, using energy that could have gone to propulsion. The 4680 format is still early in volume production, but the direction is clear: Tesla is betting that manufacturing innovation can close the energy-density gap while keeping costs low enough to support both consumer cars and high-duty-cycle robotaxis.

Different regions emphasize different metrics. In China the priority is cost per kilowatt-hour and supply-chain independence. In Europe the focus is on recyclability and overall carbon footprint of the battery’s full life. In North America the conversation still circles back to range and fast-charging speed because distances are larger and charging networks are patchier. Tesla’s decision to offer both chemistries in different models shows they’re reading these regional signals rather than forcing one universal solution.

Ultimately chemistry is not a winner-take-all contest. It’s a set of engineering trade-offs that interact with vehicle architecture, duty cycle, and local energy prices. The companies that treat it as a flexible tool rather than a single bet are the ones best positioned as the market fragments into different use cases — commuter cars, long-haul trucks, delivery vans, and autonomous fleets. Tesla’s 4680 effort is their attempt to rewrite the cost-density curve instead of simply choosing from the menu everyone else is using.

Full Self-Driving Architecture – Explained Like You Just Bought Your First Tesla

Imagine you just picked up your first Tesla and you’re wondering how the car actually sees the world and decides what to do. Let’s walk through it simply, the way I wish someone had explained it to me.

Start with the hardware journey. Early cars used HW2 and HW2.5 — basically a set of cameras plus radar and ultrasonic sensors. The idea was to give the car multiple ways of understanding distance and speed, the same way humans use sight and hearing. Then Tesla made a bold call: drop the radar and go vision-only with HW3. The bet was that cameras plus enough computing power and smart software could understand the world better than mixing sensors that sometimes contradict each other.

HW3 cars got a powerful onboard computer designed specifically for neural networks. Later HW4 added more cameras, better resolution, and a faster processor. The newest hardware is even more capable, but the important shift wasn’t just the chips — it was the software philosophy.

Tesla feeds millions of miles of real driving data into massive training clusters. The system uses something called an occupancy network. Think of it as the car building a 3D model of everything around it — not just “there’s a car” but “there’s a car occupying this exact space, moving this direction, and it’s likely to keep doing that.” Then a transformer-based planning system (the same kind of tech behind chatbots but for driving) looks at that 3D scene and figures out the safest, most efficient path forward.

To a new owner it can feel like magic when the car smoothly changes lanes or stops for a pedestrian who steps out unexpectedly. Under the hood it’s the result of iterating on vision-first perception, using real-world data as the teacher, and continuously pushing updates over the air. Unlike some competitors who rely on pre-mapped routes or expensive lidar sensors, Tesla’s approach is meant to work anywhere a human can drive, using mostly the same cameras your eyes use.

Right now the system is still supervised — you have to pay attention and be ready to take over. That’s the honest state in 2026. The leap to unsupervised robotaxi operation will require proving to regulators and to customers that the edge cases are handled reliably enough. Tesla’s advantage is the sheer volume of data and the ability to improve every car at once. Their disadvantage is that they’re doing it in public, so every mistake makes headlines.

For a new investor or owner, the simplest way to think about it is this: Tesla is treating self-driving like a software problem more than a hardware one. They bet that improving the neural nets with real-world miles will get them there faster and cheaper than bolting expensive sensors on every vehicle. Whether that bet pays off is still unfolding, but the architecture they’ve chosen is elegant in its simplicity — cameras, neural nets, and continuous learning. It’s the same first-principles mindset they apply to batteries and manufacturing: strip away the unnecessary parts and solve the problem at its core.

The two deep dives show how connected everything at Tesla really is. Better batteries make robotaxis more economical. Smarter self-driving software makes the expensive hardware work harder. It’s all one system, even if the headlines focus on one piece at a time.

Talk soon — keep an eye on how those FSD subscriptions trend, I have a feeling that number is going to tell us a lot about what’s actually happening on the road.

Sources

news.google.com