THE APEX TIMES
The vehicle-AI provider says its “AI Box” uses on-device inference to support language-model driven experiences, pairing NVIDIA’s Nemotron with ByteDance’s Doubao.
PATEO, a company focused on in-vehicle artificial intelligence, says it is moving beyond cloud-only assistants by deploying a pair of large language model tools directly on an in-car computing unit. In its latest push, the system combines NVIDIA Nemotron and ByteDance’s Doubao, according to a report citing the rollout.
Central to the effort is PATEO’s AI Box, which the company says is built on NVIDIA’s DRIVE AGX Thor platform. DRIVE AGX Thor is NVIDIA’s automotive supercomputing system designed to run advanced AI workloads in vehicles, including perception, planning, and other forms of real-time inference at the edge rather than relying on remote servers.
PATEO’s approach highlights a key distinction in the vehicle-AI market: local inference. The report describes the solution as using on-device processing, which can reduce latency and keep certain language-model capabilities available even without a strong network connection. In this design, the language models are intended to run as part of the vehicle’s onboard intelligence stack.
On the model side, NVIDIA Nemotron is described in the report as one component of the system, paired with ByteDance’s Doubao, which is characterized as a 35-billion-parameter language model. A language model with tens of billions of parameters is designed to generate and interpret text, and companies pursuing automotive deployments typically look for it to power conversational interfaces, vehicle understanding, and natural-language control of in-car functions.
While the report links the solution to both NVIDIA and ByteDance models, it does not lay out detailed engineering specifics such as optimization methods, compression or quantization strategies, or how the models are orchestrated inside the AI Box. It also does not describe the exact user-facing features, such as whether the AI is focused on infotainment, driver assistance, or operations for fleet and logistics vehicles.
NVIDIA’s role in the program is presented through the underlying hardware platform. DRIVE AGX Thor is positioned as the compute foundation for PATEO’s AI Box, implying that the platform is expected to handle the throughput and real-time constraints that in-vehicle deployment requires.
For NVIDIA, partnerships like this reflect a broader strategy in automotive AI, where the company sells both the compute platforms and the software ecosystem used to run AI models at the edge. For vehicle makers and suppliers, the promise is a more consistent path from model development to deployment, without depending on continuous connectivity.
Even so, the publicly available information in the report is limited. PATEO and NVIDIA do not disclose in the cited article what specific models are running, at what performance targets, in which vehicle programs, or what the rollout timeline and geographic coverage are. As a result, it remains unclear how widely the AI Box has been deployed and what measurable outcomes, such as response times or driver-task improvements, the companies are targeting or have achieved.
In an earnings call summary carried by Yahoo Finance, Apple said it finished a record fiscal year and posted a September-quarter record, while flagging margin pressure tied to soaring memory costs and a decelerating Services segment.
The Alphabet unit says it is funding NUS and NTU research and capacity building tied to AI for science, computing efficiency, and AI safety, alongside additional support for other research initiatives in Singapore.
Investor sentiment in Asia was split as traders assessed Apple’s latest earnings outlook while also positioning ahead of the U.S. midterm elections and digesting a fresh policy move from Australia’s central bank.
On CNBC’s Jim Cramer, speaking as shares of Apple and other tech names slid, pushed back on what he called unusually negative sentiment, arguing that it is still not enough to match the market’s downturn.
In its latest earnings discussion, Palantir said commercial revenue surpassed government revenue for the first time, underscoring a long-running push to expand beyond defense and intelligence customers.
Alphabet’s Google is expanding mobile defenses, threat-intelligence sharing, and community education efforts across East Asia, Southeast Asia, Australia, and New Zealand as scam losses climb.
A shift in how much the US government is issuing in Treasurys, and what that means for longer-term interest rates, is feeding into valuation debates for high-growth equities including NVIDIA, according to a market report.
Oracle set terms for a concurrent underwritten public offering that includes common stock sold at $116.00 per share, as the company seeks fresh capital alongside issuance of senior notes.
Traders weighed strong iPhone sales against higher memory-chip costs, sending some Apple supply-chain names lower in South Korea even as certain Chinese manufacturers advanced.
Palantir CEO Alex Karp told investors that customers are increasingly seeing value from the company’s AI-driven data analytics platform, a message framed as “surging” interest following the latest quarterly results.