Tesla Patent Reveals How Tesla Optimizes FSD

January 7, 2025

By Karan Singh

Not a Tesla App

As a continuation of our series on Tesla’s patents, we’re taking a look at how Tesla optimizes the performance of AI - FSD, in this case - in autonomous vehicles and robots. Patent WO2024073115A1 goes over efficiently running complex AI models on specialized hardware.

Before we dive into this article, we recommend reading our article on How FSD Works and our other article on Tesla’s Universal Translator for streamlining FSD deployments. While they’re not necessary, the background knowledge will help you appreciate all the details behind how Tesla does their optimization.

Just like before, we’ll be breaking this down into sections and making them as easily understandable as possible.

AI Subnetworks

FSD isn’t a monolithic entity - it is composed of smaller, specialized sub-networks, each dedicated to a specific aspect or function of autonomous operation. This modular design means that Tesla can work on improving one or all sections through training. When one section is improved, the end-to-end nature of the AI also means that the other sections will learn to adapt to the improvements and, therefore, perform better. It also allows for more efficient processing and adaptability during deployment and initial platform training.

These sub-networks might be responsible for tasks such as:

Recognizing and interpreting traffic signals
Detecting and tracking moving objects including vehicles, pedestrians, cyclists, and more
Maintaining lane position and navigating roads
Generating 3D maps of the surrounding environment
Planning paths and making real-time driving decisions

This division of labor allows FSD to handle the complexities of autonomous driving with greater efficiency and precision

Tailored Compilers

Different hardware components are good at different things - and they also require different types of instructions. CPUs, GPUs, and specialized AI accelerators (NPUs) all have unique architecture and capabilities.

Tesla uses a compiler toolchain to translate FSD into machine code that is specifically tailored to each hardware component. This ensures that instructions are executed optimally on each processor, maximizing performance and efficiency.

Strategic Assignment

To further optimize performance, Tesla employs a system that intelligently assigns each FSD sub-network to the most suitable hardware component. This ensures that computationally demanding tasks are handled by the most powerful processors while simpler tasks are delegated to more efficient units.

This strategic assignment of tasks maximizes the overall efficiency of the system, ensuring that each component operates within its optimal performance range.

Optimized Scheduling

The order in which the hardware executes instructions also plays a crucial role in performance. Tesla's system includes an "execution scheduler" that determines the most efficient sequence of operations, minimizing delays and maximizing real-time responsiveness.

This optimized scheduling ensures that the FSD can react quickly and make informed decisions in dynamic driving situations - or quick-response situations with Optimus - like catching a ball.

While the demo here has been confirmed to be teleoperated, Tesla has said they’re working to let Optimus do this autonomously in the future.

Let’s give Optimus a hand for catching ball!
pic.twitter.com/i44qcD1iLd
— Elon Musk (@elonmusk) November 28, 2024

Quantization-Aware Training

To reduce the computational burden and power consumption of FSD, Tesla employs a technique called "quantization-aware training." This involves training FSD to work with lower-precision numbers, which require less processing power and memory. Essentially - rounding.

This approach allows the AI to operate efficiently without significantly compromising accuracy, striking a balance between performance and resource utilization.

Clock Synchronization

In hardware systems with multiple chips, maintaining precise timing is crucial for accurate and synchronized operation. Tesla's system incorporates mechanisms to synchronize the clocks of all processing units, preventing timing errors and ensuring seamless coordination between different components.

This precise clock synchronization is essential for FSD to make accurate real-time calculations and respond effectively to changing conditions.

Redundancy and Failover

To ensure reliability and safety, Tesla's system supports redundant hardware configurations. This means that if a critical component fails, a backup component can seamlessly take over, preventing disruptions in operation.

This redundancy and failover capability is crucial for maintaining the safety and integrity of autonomous systems, especially when driving. Tesla has built-in both physical and software redundancy to FSD, ensuring that it maintains a minimum standard of safety when operating autonomously.

In Simpler Terms…

Imagine a large company (FSD) with different departments (sub-networks) responsible for specific tasks. Each department has its own specialized tools and equipment (hardware components). Tesla's system acts like an efficient management structure, assigning the right tasks to the right departments, providing them with the appropriate tools, and coordinating their efforts for optimal productivity and performance.

New Castings Spotted at Giga Texas Likely Intended for Tesla Cybercab

April 21, 2025

By Karan Singh

@JoeTegtmeyer

Tesla’s Giga Texas factory usually gives us the first site of Tesla’s upcoming products. We first saw the Cybertruck and Model Y castings here. With Giga Texas being one of Tesla’s largest factories, it’s logical that most products would originate here.

Tesla has also stated that it intends to manufacture the Cybercab, Semi, the next-generation vehicle, and Optimus at Giga Texas over the coming years. The affordable vehicle and Cybercab were originally intended to be manufactured at Giga Mexico, but the plans for that facility were waylaid by changes in economic policy.

Robotaxi Castings

These new castings were spotted by Joe Tegtmeyer, who regularly does drone flights of Giga Texas. Joe pointed out that these castings don’t look like the usual Model Y or Cybertruck castings usually seen outside Giga Texas.

With an eagle eye, @minusYCore on X also spotted some interesting text on the frames holding the castings up. In particular, the castings say “RTTX050” and “W68-RSF AS-CAST”. These could be interpreted as ‘Robotaxi Texas’ and ‘Rear SubFrame.’ The as-cast portion indicates that these particular castings haven’t been trimmed yet, according to the X user.

The castings laid out.

@JoeTegtmeyer

The size and shape of these castings—combined with rumors that Tesla’s more affordable vehicle has been delayed—suggest they are intended for the Cybercab.

These castings are much flatter and appear to be a different size than the castings found elsewhere throughout Giga Texas, indicating that they are intended for a different product entirely.

It’s possible that these are the first castings used by Tesla to test their unboxed assembly process, which the Cybercab is expected to rely on. If you take a closer look at the video below, you’ll note that these new castings look very similar to the ones in the unboxed assembly video.

Interestingly, Tesla did say that they don’t intend to have the Cybercab available for customers before late 2026 or early 2027, but we’ll likely hear updated timelines as Tesla’s Q1 2025 Earnings Call tomorrow.

A more vertical look at the castings.

@JoeTegtmeyer

New Giga Presses

To top it all off, new parts for a Giga Press - the machine Tesla uses to make these castings - were also sighted in Texas. These machines are few and far between, and each one is highly specialized for the particular vehicle it produces. Seeing new parts coming in usually indicates that a new assembly line is under construction, or that changes are being made to an existing line to either expand it or update it.

There’s a lot happening and we will hopefully know more tomorrow evening.

New Giga Press parts

@JoeTegtmeyer

PepsiCo Building Additional Tesla Semi Chargers

April 21, 2025

By Karan Singh

Not a Tesla App

Tesla’s first large-scale partner for the Tesla Semi - PepsiCo - is building out another Semi Megacharger site per a recent filing found by MarcoRP. Building out these Semi charging sites is expensive due to their much higher power output when compared to Superchargers.

Name Change

Interestingly enough, Tesla appears to have shifted its terminology recently, now often referring to the Semi’s high-power charging solution as Tesla Semi Chargers instead of the previously common "Megachargers." Regardless of the name, these chargers are crucial for the Semi. They’re designed to deliver 1.2 megawatts of charging power capable of adding hundreds of miles of range in roughly 30 minutes.

Frito-Lay Site

The new site detailed in the recent filing will be located at a Frito-Lay production plant on Hageman Road in Bakersfield, California. Permits were filed just a couple of days ago for the installation of eight Tesla Semi Charger stalls along with their associated electrical equipment.

This facility adds to PepsiCo's existing network of chargers, supporting the Tesla Semis operating out of their Sacramento facility, and highlights Bakersfield as the next expansion of the Semi’s operational area. Notably, this new Frito-Lay charging site is strategically located just five miles (approx. 8 km) away from a massive future public charging complex Tesla is planning on Kelton Way.

Kelton Way Charging Complex

Tesla’s planned charging complex in Bakersfield, which is just down the road, will feature 72 Supercharger stalls, as well as 18 Semi charger stalls. According to pre-permits filed last month, this site will be fairly large and accessible to both the public and Tesla’s partners.

A rendering of the Kelton Way site - with the Semi Chargers on the outside, and the Covered Supercharger site in the center.

@MarcoRPi1 on X

Tesla plans to develop a vacant plot right off the interstate into a large charging hub, which is now the second major public Semi Charger site we’ve heard about. For now, these sites will remain rare, but we expect that Tesla and its partners will be happy to expand them out further as required.

Carson Semi Charger Site

These Bakersfield developments – both the private Frito-Lay site and the large public Kelton Way hub will help lay out the necssary infrastructure for the Tesla Semi, much like the early days of Tesla’s Superchargers. Since Semi cannot be charged at a Supercharger site (the MCS plug is notably different than NACS), these sites are a must-have.

Tesla's first publicly accessible Semi Charger site planned near the Port of Long Beach in Carson, California, was one of the first signs of these types of Semi Charger facilities being established - and like Kelton Way, it also includes amenities for truck drivers. Tesla is also considering what long, medium, and short-haul truckers need for their stays at these sites.

As Tesla works to ramp up production of the Semi itself, the parallel expansion of dedicated, high-power charging infrastructure by both Tesla and its key partners are essential to making electric heavy-duty trucking a reality.