For years, Elon Musk has talked about Dojo — the AI supercomputer that would be the cornerstone of Tesla’s AI ambitions. It’s necessary sufficient to Musk that in July 2024, he stated the corporate’s AI crew would “double down” on Dojo within the lead-up to Tesla’s robotaxi reveal, which occurred in October.
But what precisely is Dojo? And why is it so essential to Tesla’s long-term technique?
In quick: Dojo is Tesla’s custom-built supercomputer that’s designed to coach its “Full Self-Driving” neural networks. Beefing up Dojo goes hand-in-hand with Tesla’s aim to succeed in full self-driving and convey a robotaxi to market. FSD, which is on a whole bunch of hundreds of Tesla automobiles at present, can carry out some automated driving duties however nonetheless requires a human to be attentive behind the wheel.
Tesla’s Cybercab reveal has come and gone, and now the corporate is gearing as much as launch an autonomous ride-hail service utilizing its personal fleet of automobiles in Austin this June. Tesla additionally stated throughout its 2024 fourth-quarter and full-year earnings name on the finish of January that it plans to launch unsupervised FSD for U.S. clients in 2025.
Musk’s earlier rhetoric has been that Dojo can be the important thing to reaching Tesla’s aim of full self-driving. Now that Tesla seems near nearing that aim, Musk has been mum on Dojo.
Instead, ever since August 2024, discuss has been round Cortex, Tesla’s “big new AI coaching supercluster being constructed at Tesla HQ in Austin to resolve real-world AI.” Musk has additionally stated it is going to have “huge storage for video coaching of FSD & Optimus.”
In Tesla’s This fall shareholder deck, the corporate shared updates on Cortex, however nothing on Dojo.
Tesla has positioned itself to spend huge on AI and Dojo — and now Cortex — to succeed in its aim of autonomy for each automobiles and humanoid robots. And Tesla’s future success actually hinges on its skill to nail this down, given the elevated competitors within the EV market. So it’s price taking a more in-depth have a look at Dojo, Cortex, and the place all of it stands at present.
Tesla’s Dojo backstory
Musk doesn’t need Tesla to be simply an automaker, or perhaps a purveyor of photo voltaic panels and power storage methods. Instead, he needs Tesla to be an AI firm, one which has cracked the code to self-driving automobiles by mimicking human notion.
Most different firms constructing autonomous car know-how depend on a mixture of sensors to understand the world — like lidar, radar and cameras — in addition to high-definition maps to localize the car. Tesla believes it could obtain absolutely autonomous driving by counting on cameras alone to seize visible knowledge after which use superior neural networks to course of that knowledge and make fast selections about how the automobile ought to behave.
As Tesla’s former head of AI, Andrej Karpathy, stated on the automaker’s first AI Day in 2021, the corporate is mainly making an attempt to construct “an artificial animal from the bottom up.” (Musk had been teasing Dojo since 2019, however Tesla formally introduced it at AI Day.)
Companies like Alphabet’s Waymo have commercialized Level 4 autonomous automobiles — which the SAE defines as a system that may drive itself with out the necessity for human intervention underneath sure circumstances — by means of a extra conventional sensor and machine studying method. Tesla has nonetheless but to provide an autonomous system that doesn’t require a human behind the wheel.
About 1.8 million individuals have paid the hefty subscription value for Tesla’s FSD, which presently prices $8,000 and has been priced as excessive as $15,000. The pitch is that Dojo-trained AI software program will finally be pushed out to Tesla clients by way of over-the-air updates. The scale of FSD additionally means Tesla has been in a position to rake in thousands and thousands of miles price of video footage that it makes use of to coach FSD. The thought there’s that the extra knowledge Tesla can gather, the nearer the automaker can get to truly reaching full self-driving.
However, some trade consultants say there is likely to be a restrict to the brute pressure method of throwing extra knowledge at a mannequin and anticipating it to get smarter.
“First of all, there’s an financial constraint, and shortly it is going to simply get too costly to do this,” Anand Raghunathan, Purdue University’s Silicon Valley professor {of electrical} and laptop engineering, informed TechCrunch. Further, he stated, “Some individuals declare that we would really run out of significant knowledge to coach the fashions on. More knowledge doesn’t essentially imply extra info, so it is dependent upon whether or not that knowledge has info that’s helpful to create a greater mannequin, and if the coaching course of is ready to really distill that info into a greater mannequin.”
Raghunathan stated regardless of these doubts, the pattern of extra knowledge seems to be right here for the short-term at the very least. And extra knowledge means extra compute energy wanted to retailer and course of all of it to coach Tesla’s AI fashions. That is the place Dojo, the supercomputer, is available in.
What is a supercomputer?
Dojo is Tesla’s supercomputer system that’s designed to operate as a coaching floor for AI, particularly FSD. The identify is a nod to the area the place martial arts are practiced.
A supercomputer is made up of hundreds of smaller computer systems referred to as nodes. Each of these nodes has its personal CPU (central processing unit) and GPU (graphics processing unit). The former handles total administration of the node, and the latter does the complicated stuff, like splitting duties into a number of elements and dealing on them concurrently. GPUs are important for machine studying operations like people who energy FSD coaching in simulation. They additionally energy massive language fashions, which is why the rise of generative AI has made Nvidia essentially the most worthwhile firm on the planet.
Even Tesla buys Nvidia GPUs to coach its AI (extra on that later).
Why does Tesla want a supercomputer?
Tesla’s vision-only method is the principle motive Tesla wants a supercomputer. The neural networks behind FSD are educated on huge quantities of driving knowledge to acknowledge and classify objects across the car after which make driving selections. That implies that when FSD is engaged, the neural nets have to gather and course of visible knowledge constantly at speeds that match the depth and velocity recognition capabilities of a human.
In different phrases, Tesla means to create a digital duplicate of the human visible cortex and mind operate.
To get there, Tesla must retailer and course of all of the video knowledge collected from its automobiles around the globe and run thousands and thousands of simulations to coach its mannequin on the info.
Tesla seems to depend on Nvidia to energy its present Dojo coaching laptop, nevertheless it doesn’t wish to have all its eggs in a single basket — not least as a result of Nvidia chips are costly. Tesla additionally hopes to make one thing higher that will increase bandwidth and reduces latencies. That’s why the automaker’s AI division determined to give you its personal {custom} {hardware} program that goals to coach AI fashions extra effectively than conventional methods.
At that program’s core is Tesla’s proprietary D1 chips, which the corporate says are optimized for AI workloads.
Tell me extra about these chips
Tesla is of an identical opinion to Apple in that it believes {hardware} and software program needs to be designed to work collectively. That’s why Tesla is working to maneuver away from the usual GPU {hardware} and design its personal chips to energy Dojo.
Tesla unveiled its D1 chip, a silicon sq. the scale of a palm, on AI Day in 2021. The D1 chip entered into manufacturing as of at the very least May this 12 months. The Taiwan Semiconductor Manufacturing Company (TSMC) is manufacturing the chips utilizing 7 nanometer semiconductor nodes. The D1 has 50 billion transistors and a big die dimension of 645 millimeters squared, in keeping with Tesla. This is all to say that the D1 guarantees to be extraordinarily highly effective and environment friendly and to deal with complicated duties rapidly.
“We can do compute and knowledge transfers concurrently, and our {custom} ISA, which is the instruction set structure, is absolutely optimized for machine studying workloads,” stated Ganesh Venkataramanan, former senior director of Autopilot {hardware}, at Tesla’s 2021 AI Day. “This is a pure machine studying.”
The D1 continues to be not as highly effective as Nvidia’s A100 chip, although, which can also be manufactured by TSMC utilizing a 7 nanometer course of. The A100 comprises 54 billion transistors and has a die dimension of 826 sq. millimeters, so it performs barely higher than Tesla’s D1.
To get the next bandwidth and better compute energy, Tesla’s AI crew fused 25 D1 chips collectively into one tile to operate as a unified laptop system. Each tile has a compute energy of 9 petaflops and 36 terabytes per second of bandwidth, and comprises all of the {hardware} needed for energy, cooling and knowledge switch. You can consider the tile as a self-sufficient laptop made up of 25 smaller computer systems. Six of these tiles make up one rack, and two racks make up a cupboard. Ten cupboards make up an ExaPOD. At AI Day 2022, Tesla stated Dojo would scale by deploying a number of ExaPODs. All of this collectively makes up the supercomputer.
Tesla can also be engaged on a next-gen D2 chip that goals to resolve info move bottlenecks. Instead of connecting the person chips, the D2 would put your entire Dojo tile onto a single wafer of silicon.
Tesla hasn’t confirmed what number of D1 chips it has ordered or expects to obtain. The firm additionally hasn’t supplied a timeline for the way lengthy it is going to take to get Dojo supercomputers operating on D1 chips.
In response to a June submit on X that stated: “Elon is constructing an enormous GPU cooler in Texas,” Musk replied that Tesla was aiming for “half Tesla AI {hardware}, half Nvidia/different” over the following 18 months or so. The “different” might be AMD chips, per Musk’s remark in January.
What does Dojo imply for Tesla?
Taking management of its personal chip manufacturing implies that Tesla would possibly at some point be capable to rapidly add massive quantities of compute energy to AI coaching packages at a low value, significantly as Tesla and TSMC scale up chip manufacturing.
It additionally implies that Tesla could not need to depend on Nvidia’s chips sooner or later, that are more and more costly and laborious to safe.
During Tesla’s second-quarter earnings name, Musk stated that demand for Nvidia {hardware} is “so excessive that it’s typically tough to get the GPUs.” He stated he was “fairly involved about really having the ability to get regular GPUs once we need them, and I feel this due to this fact requires that we put much more effort on Dojo so as to be certain that we’ve obtained the coaching functionality that we want.”
That stated, Tesla continues to be shopping for Nvidia chips at present to coach its AI. In June, Musk posted on X:
Of the roughly $10B in AI-related expenditures I stated Tesla would make this 12 months, about half is inside, primarily the Tesla-designed AI inference laptop and sensors current in all of our automobiles, plus Dojo. For constructing the AI coaching superclusters, Nvidia {hardware} is about 2/3 of the price. My present greatest guess for Nvidia purchases by Tesla are $3B to $4B this 12 months.
“Inference compute” refers back to the AI computations carried out by Tesla automobiles in actual time and is separate from the coaching compute that Dojo is chargeable for.
Dojo is a dangerous guess, one which Musk has hedged a number of occasions by saying that Tesla won’t succeed.
In the long term, Tesla might theoretically create a brand new enterprise mannequin primarily based on its AI division. Musk has stated that the primary model of Dojo can be tailor-made for Tesla laptop imaginative and prescient labeling and coaching, which is nice for FSD and for coaching Optimus, Tesla’s humanoid robotic. But it wouldn’t be helpful for a lot else.
Musk has stated that future variations of Dojo can be extra tailor-made to general-purpose AI coaching. One potential downside with that’s nearly all AI software program out there was written to work with GPUs. Using Dojo to coach general-purpose AI fashions would require rewriting the software program.
That is, until Tesla rents out its compute, just like how AWS and Azure hire out cloud computing capabilities. Musk additionally famous throughout Q2 earnings that he sees “a path to being aggressive with Nvidia with Dojo.”
A September 2023 report from Morgan Stanley predicted that Dojo might add $500 billion to Tesla’s market worth by unlocking new income streams within the type of robotaxis and software program providers.
In quick, Dojo’s chips are an insurance coverage coverage for the automaker, however one that would pay dividends.
How far alongside is Dojo?
Reuters reported final 12 months that Tesla started manufacturing on Dojo in July 2023, however a June 2023 submit from Musk prompt that Dojo had been “on-line and operating helpful duties for just a few months.”
Around the identical time, Tesla stated it anticipated Dojo to be one of many high 5 strongest supercomputers by February 2024 — a feat that has but to be publicly disclosed, leaving us uncertain that it has occurred.
The firm additionally stated it expects Dojo’s whole compute to succeed in 100 exaflops in October 2024. (One exaflops is the same as 1 quintillion laptop operations per second. To attain 100 exaflops, and assuming that one D1 can obtain 362 teraflops, Tesla would want greater than 276,000 D1s, or round 320,500 Nvidia A100 GPUs.)
Tesla additionally pledged in January 2024 to spend $500 million to construct a Dojo supercomputer at its gigafactory in Buffalo, New York.
In May 2024, Musk famous that the rear portion of Tesla’s Austin gigafactory can be reserved for a “tremendous dense, water-cooled supercomputer cluster.” Now we all know that it’s really Cortex, not Dojo, that’s taking on that area in Austin.
Just after Tesla’s second-quarter earnings name, Musk posted on X that the automaker’s AI crew is utilizing Tesla HW4 AI laptop (renamed AI4), which is the {hardware} that lives on Tesla automobiles, within the coaching loop with Nvidia GPUs. He famous that the breakdown is roughly 90,000 Nvidia H100s plus 40,000 AI4 computer systems.
“And Dojo 1 could have roughly 8k H100-equivalent of coaching on-line by finish of 12 months,” he continued. “Not huge, however not trivial both.”
Tesla hasn’t supplied updates as as to if it has gotten these chips on-line and operating Dojo. During the corporate’s fourth-quarter 2024 earnings name, nobody talked about Dojo. However, Tesla stated it accomplished the deployment of Cortex in This fall and that it was Cortex that helped allow V13 of supervised FSD.
This story initially revealed August 3, 2024, and we’ll replace it as new info develops.
