NextFin News - Koji Sato, the executive who remade Toyota Motor Corp. as its president and now steers its industry strategy as vice chairman and chief industry officer, says artificial intelligence can accelerate autonomous driving. The comment, made in a September 4, 2026 interview, is more than a technology endorsement: it is the world's largest automaker declaring that the race for self-driving cars will be decided by who owns the data, not just who wrote the code first.
That claim lands at a moment when Toyota is simultaneously the most profitable carmaker on earth and the company investors most often accuse of moving too slowly on software. For the fiscal year ended March 2026, Toyota reported consolidated net revenues of 50.68 trillion yen and operating income of 3.77 trillion yen, with net income attributable to the company of 3.85 trillion yen. In the quarter ended June 2026, net profit jumped 76% to 1.48 trillion yen, and management raised its full-year revenue forecast to 54 trillion yen while approving a share buyback of about 1 trillion yen. Yet Toyota's shares trade near $195, down from a 52-week high of $248.90 reached in February, at a price-to-earnings ratio of roughly nine — a valuation that prices the company as a hardware manufacturer, not a software platform.
Sato's argument is that AI changes the economics of autonomy enough to close the gap. Toyota is backing that view with a ¥500 billion ($3.3 billion) joint investment with Nippon Telegraph & Telephone Corp. through 2030, a plan to bring a Level 2++ driver-assistance system to passenger vehicles from 2028, and an exploration of a deeper partnership with Alphabet's Waymo. The question for investors is whether this is a credible catch-up strategy or a four-year-late admission that Toyota missed the first act of the software-defined vehicle era.
The Bet: Data, Not Just Code, Decides Autonomy
The core of Sato's position is a shift in what autonomy requires. For most of the 2010s, the industry treated self-driving as a mapping and sensor problem: build high-definition maps, stack lidar and radar on the roof, and hand-code rules for every scenario. That approach produced working robotaxis, but only inside geofenced cities, at a capital intensity that never scaled. Waymo, the leader, had logged 127 million miles of autonomous operation and reached 500,000 paid weekly trips across 10 U.S. cities by March 2026 — impressive, but a fraction of the trillions of miles driven globally each year.
AI-driven autonomy works differently. Instead of coding rules, companies train neural networks on fleet data; the system generalizes to new roads, and every mile driven makes it better. The moat becomes data volume and quality, compounded over time. Tesla is the purest expression of this model: 1.48 million active Full Self-Driving subscriptions as of the second quarter of 2026, up 56% year over year, on a fleet that surpassed 10 billion cumulative supervised miles in early May 2026. Baidu's Apollo Go, meanwhile, reported 20 million cumulative rides worldwide by February 2026, showing that Chinese operators are building their own data flywheel.
Toyota's counter is that it already sits on an underused data asset. The company sells roughly 10 million vehicles a year — consolidated sales reached 9.59 million units in fiscal 2026, with total group sales setting a record of 11.28 million. A century of real-world driving experience is embedded in its engineering. And with NTT, it is building a network that transmits vehicle data at scale — the infrastructure piece that turns parked cars into a distributed sensor array. At the briefing where the NTT partnership was announced, Sato framed the challenge in one line:
"Transmitting large amounts of data will be crucial as software-defined vehicles become more common."
The mechanism Sato is describing is compounding: vehicles generate data, AI improves the model, the improved model makes the feature more valuable, more customers adopt it, and the fleet generates more data. This is a structural shift, not a cyclical fluctuation. The move from hand-coded rules to data-trained neural networks changes the cost curve permanently: marginal miles approach zero cost, and a leader's advantage compounds rather than mean-reverts. Cyclical factors — chip supply, interest rates, quarterly deliveries — will still drive Toyota's earnings from quarter to quarter. But the autonomy race itself does not revert; once a data moat forms, it does not erode on its own. That is why being four years late matters more than a four-year earnings cycle would suggest.
The weakness in Toyota's argument is equally clear. Tesla's data is FSD-engaged miles — every intervention is a labeled training example. Most of Toyota's fleet data is human-driven, which is useful but not equivalent. Toyota must either collect its own edge cases at scale or synthesize them in simulation. That is the gap the ¥500 billion is meant to close.
The Strategy: Build Level 4, Sell Level 2++
Toyota's commercial posture is deliberately conservative, and that conservatism is the most interesting part of the strategy. The company plans to bring a Level 2++ driver-assistance system to passenger vehicles beginning in 2028, expanding across more models from 2030. Akihiro Sarada, head of Toyota's Software Development Center, put the logic plainly:
"We'll develop Level 4 technology from a technical standpoint but offer it commercially as Level 2++."
Read that sentence twice. Toyota intends to build technology capable of greater autonomy than customers will initially receive. Under a Level 2++ framework, the driver remains responsible, the regulatory burden stays light, and the feature can be sold as software on millions of privately owned cars. That avoids the liability, permitting, and public-trust hurdles that have slowed fully driverless deployment. It also converts Toyota's greatest asset — a global installed base of roughly 10 million vehicles a year — into a distribution channel for recurring software revenue.
Commercial vehicles are moving faster. Toyota has said it aims to bring Level 4 capability to its e-Palette mobility vehicle during fiscal 2027, and the company is exploring a broader autonomous-driving partnership with Waymo that could eventually bring Waymo technology into personally owned Toyota vehicles. The two tracks are complementary: e-Palette proves the stack in controlled commercial use, while the passenger-car strategy monetizes autonomy as a feature before the regulatory environment permits unsupervised operation.
The risk is timing. Toyota's 2028 passenger-car launch comes after competitors have already moved. Mercedes-Benz debuted MB.DRIVE ASSIST PRO at CES 2026 and launched it in China at the end of 2025, replacing Drive Pilot with a Level 2++ system that guides vehicles through city streets under driver supervision. General Motors' Super Cruise and Ford's BlueCruise already offer supervised hands-free highway driving. Stellantis partnered with Wayve in May 2026, targeting a 2028 launch of AI-powered supervised hands-free systems — the same year Toyota plans to arrive. In a feature race, being first matters more than being best.
Toyota is not starting from zero. Teammate, its advanced driver-assistance technology, launched in Japan in April 2021 on the Lexus LS and Mirai; its Advanced Drive function enables automatic steering, acceleration, and braking on highways under active driver supervision, and is classified as an SAE Level 2 system. The company's joint venture with Pony.ai transitioned to mass production of bZ4X-based robotaxis in February 2026, targeting more than 1,000 units for commercial service in China. And in February 2026, Toyota signed a Robots-as-a-Service agreement with Agility Robotics to deploy seven Digit humanoid robots for logistics at its Woodstock, Ontario plant. These are real deployments, but they are pilots next to the scale Toyota needs for an AI flywheel.
The Money: Record Profits Fund a Late Entry
Toyota can afford to be late because it is rich. The company's AI investment stack has grown from scattered pilots to a coordinated offensive. In September 2025, Toyota launched two funds totaling $1.5 billion: Toyota Invention Partners, with ¥100 billion (about $670 million) in capital for early-stage startups, and Woven Capital Fund II, a ¥120 billion (about $800 million) vehicle for growth-stage companies in AI, automation, and energy. In November 2025, it announced an additional investment of up to $10 billion in U.S. operations over the next five years, bringing total U.S. investment to nearly $60 billion. These sit alongside a broader $11.2 billion investment plan for growth technologies, including AI and EVs, announced in May 2024 after the company posted record profits.
The domestic market is also shifting beneath Toyota's feet. Japan's automotive software market is projected to grow from approximately $5.8 billion in 2025 to nearly $13 billion by 2030, according to the Yano Research Institute, with software-defined-vehicle and AI-driven solutions accounting for more than half of total market value. Toyota is not chasing a niche; it is chasing a market that is set to more than double in five years.
And the financial results give management room to spend. In the first quarter of fiscal 2027, Toyota raised its full-year revenue forecast to 54 trillion yen, 6.5% above fiscal 2026, and lifted its net profit forecast to 3.25 trillion yen — still 15.5% below the prior year, but supported by a weaker yen, higher financial income, and solid demand in the U.S. and India. Cash and cash equivalents stood at 10.3 trillion yen at the end of June 2026. The company also approved a share buyback of about 1 trillion yen, roughly $6 billion.
The Counter-Thesis: Scale Belongs to the Operators
The strongest argument against Toyota's position is that autonomy is won by whoever owns the miles loop, not whoever ships the most cars. Waymo's 500,000 weekly paid trips and Tesla's more than 10 billion supervised miles compound in ways that a 2028 launch cannot retroactively match. By the time Toyota's Level 2++ reaches customers, Tesla's Full Self-Driving will have accumulated years more data, and Waymo will have expanded into more cities. Four years is an eternity in machine learning.
There is also a data-quality problem. Toyota's fleet data is largely human-driven. Tesla's 1.48 million paying FSD users generate intervention-labeled data on every trip — precisely the edge cases that make or break an autonomy stack. A century of engineering experience does not automatically translate into labeled training data. Toyota can simulate edge cases, but simulation is only as good as the models that generate it. This is the gap money can narrow but not erase.
Finally, there is the culture question. Toyota's competitive advantage was built on manufacturing discipline, supply-chain management, and quality control — capabilities that compound in metal, not in neural networks. The $1.5 billion in venture funds and the NTT partnership can buy talent and infrastructure, but they cannot instantly create a software-native organization. GM, Mercedes, and Hyundai have been reorganizing around software for years; Toyota is reorganizing now.
The answer to the counter-thesis is that Toyota is not trying to win the robotaxi race. It is trying to monetize autonomy as a software feature on the largest distribution base in the industry. The NTT data network — aimed at predicting and responding to traffic accidents — is a different kind of moat: infrastructure that improves road safety and creates a recurring data relationship, not just an in-car feature. And the Waymo option keeps the door open to full autonomy without betting the company on it today.
But the timing risk is real and specific. Tesla has tied real robotaxi scale to its next-generation Full Self-Driving model, expected in late 2026 or early 2027. If that rewrite delivers unsupervised driving across multiple cities at meaningful scale, Toyota's 2028 Level 2++ launch will look like yesterday's news before it reaches the market. The conservative path is safer, but safety has an opportunity cost.
What to Watch: The 2027 Proof Points
The judgment on Sato's AI bet does not require faith; it requires watching three signals.
First, the Level 2++ roadmap. Toyota must name specific models and an attach-rate target for the 2028 launch by the end of 2027. Without a concrete lineup and adoption goal, the strategy remains a press release. Second, the Waymo partnership. A signed agreement to integrate Waymo technology into personally owned Toyota vehicles would validate the two-track approach; talks that stall past 2027 without a deal would suggest Toyota is hedging rather than committing. Third, the data flywheel itself. If Toyota cannot demonstrate that its fleet data — combined with NTT's network — produces intervention rates competitive with Tesla's FSD on comparable roads, the core premise of the strategy fails.
The scenarios break down by time horizon. In the short term, Toyota's stock remains an earnings story, not a software story: hybrids, the yen, and U.S. demand drive results, and the roughly nine-times earnings multiple reflects that reality. Over the medium term, 2028 to 2030, the Level 2++ rollout is the proof point — how many models receive it, how many customers pay for it, and whether the Waymo deal closes. Over the long term, the question is structural: does Toyota's data plus the NTT network become a defensible autonomy platform, or does the company end up a feature supplier in someone else's ecosystem?
The base case is that Toyota ships Level 2++ on key models in 2028, keeps the Waymo partnership limited to commercial e-Palette vehicles, and captures meaningful software revenue without claiming autonomy leadership. The upside case is that Waymo integration into personal vehicles closes, the NTT accident-prediction network becomes an industry standard, and Toyota becomes the default platform for autonomy in privately owned cars — the Android of the road. The downside case is that 2028 slips, Tesla's next-generation system achieves unsupervised scale first, and Toyota's Level 2++ arrives as a me-too feature in a crowded lane, compressing margins without creating a moat.
Sato's AI bet is not a claim that Toyota will out-innovate Silicon Valley at its own game. It is a wager that the last mile of autonomy will be won by the company that already owns the road, not the one that wrote the code first. The next 18 months will show whether that is insight or rationalization.
Explore more exclusive insights at nextfin.ai.
