Texas Instruments (TI), regarded as a powerhouse in analog and embedded semiconductors, is delivering results in the Autonomous Driving vehicle industry.
Since its founding in 1930, TI has achieved clear results in analog chips that measure continuous signals such as voltage and temperature, and in embedded semiconductors that control functions such as sensors and communications inside devices. TI's roughly 80,000 products are used by more than 100,000 corporations. Last year's revenue was $17.682 billion (about 26.164 trillion won), 95% of which came from the analog and embedded chip business.
TI consolidated the technology it built in analog and embedded chips to develop automotive semiconductors that it launched early this year. TI said, "Customers adopting our products will be able to reach 'Autonomous Driving Level 3 (L3)' with ease," adding, "By emphasizing high performance even at low power, we have recently secured a diverse range of customers, showing results in terms of market penetration." L3 corresponds to "conditional Autonomous Driving," in which the driver intervenes only in specific situations and the vehicle drives itself in most sections.
According to market research firm Stratview Research, the Autonomous Driving semiconductor market TI has entered is expected to grow from $67.2 billion (about 100 trillion won) last year to $134.32 billion (about 199 trillion won) by 2032.
◇ "Industry-leading power efficiency"
TI's Autonomous Driving solution lineup consists of ▲ TDA5 high-performance system-on-chip (SoC) ▲ 4D imaging radar transceiver (product name AWR2188) ▲ Ethernet physical layer (internal communications that transmit data generated by Autonomous Driving vehicle sensors, product name DP83TD555J-Q1 10BASE-T1S). Each serves as the brain, eyes, and neural network in an Autonomous Driving vehicle.
Mark Ng, TI's senior director of automotive systems, cited "industry-leading power efficiency" as a differentiator for the TDA5 family in a recent written interview with ChosunBiz. "The TDA5 family is a new solution from TI, which has more than 20 years of experience in automotive processing," Ng said, adding that it was developed with a focus on scalability, safety, and efficiency. He added, "The biggest advantage of the TDA5 chip, which plays the role of the brain in Autonomous Driving vehicles, is that it can process computations quickly even with low power."
The TDA5 chip delivers up to 1,200 TOPS (1,200 trillion operations per second; 1 TOPS is 1 trillion operations per second). It supports 24 TOPS per watt (W). "Today, to achieve 400 TOPS of artificial intelligence (AI) processing, many automakers often consume more than 150 W," Ng said. "TI's chip can run 400 TOPS at under 50 W." He added, "Compared with competing solutions, it can process AI tasks faster with less power," noting that it reduces power and cooling requirements and helps customers implement advanced driver-assistance systems (ADAS) more easily.
To realize the low-power, high-performance TDA5 chip, TI integrated its C7 Neural Processing Unit (NPU). This improved AI processing performance by up to 12 times over the previous generation. The manufacturing process uses 5 nm (nanometers, one-billionth of a meter) qualified for automotive. Built on UCle (an open standard interface to efficiently link and communicate chiplets within a single package), the chip supports chiplet architecture (a technology that splits high-performance semiconductor functions into multiple parts and then combines them), enabling customized applications for customers.
"The TDA5 chip supports large language models (LLMs), vision-language models (VLMs), and advanced transformer networks with billions of parameters," Ng said, adding, "It can scale flexibly from Autonomous Driving L1 to L3."
TI also offers a virtual development kit in collaboration with Synopsys to boost customer development efficiency. With development tools that include a digital twin (a concept that precisely implements the real world in digital space for simulations), customers can test how TDA5 can be applied to their products without a vehicle. TI described it as a capability that can help shorten time to market for software-defined vehicles (SDVs) by up to 12 months.
◇ "We will do our best to support Korean customers developing Autonomous Driving technology"
Products that serve as the "eyes" and "neural network" of Autonomous Driving vehicles also gained various functions. TI applied an analog-to-digital converter (a device that converts physical signals into data) and a radar chirp signal slope engine (a technology that calculates the rate of change of signals emitted by radar to measure distance more accurately) to the new radar transceiver, improving performance by up to 30% over existing products.
This enables detection of objects more than 350 meters away. It offers resolution high enough to distinguish two vehicles in close proximity or spot small objects on the road. It provides eight transmit channels and eight receive channels (8x8) via a single LOP (a method of delivering RF signals directly from the package to the antenna through a waveguide), enabling a high-resolution radar system along with central controller processing. The company said this allows for simplified vehicle design.
"The newly launched transceiver uses 4D imaging technology to detect where an object is, how tall it is, and how quickly it is moving away or approaching," Ng said. "Many sensors struggle to measure accurately in adverse weather such as fog and rain or at night, but TI's transceiver is radar-based and operates consistently in all weather conditions." He added that the product delivers stable performance in challenging situations, such as early detection of highway hazards and detecting falling cargo or objects in high dynamic range (HDR) environments.
Signal delivery speed is considered one of the key factors in realizing Autonomous Driving vehicles. Ng offered this analogy for Ethernet's role: "Imagine you're riding a bicycle and a ball rolls in. Your eyes spot the ball and send this information to your brain. The brain, upon receiving the signal, decides to stop. It sends the command 'grab the brake' through the nerves, and your hands and feet execute it. TI's sensing, networking, and processing technologies implement this sequence in Autonomous Driving vehicles."
TI's Ethernet supports synchronization at the nanosecond level (one-billionth of a second). It boosts performance by embedding a media access controller (MAC, which manages data transmission order and collisions) into a serial peripheral interface (SPI, a communication method for sending and receiving data between chips). "The moment the radar detects an obstacle, that information is immediately delivered to the central chip (TDA5), enabling the vehicle to respond in real time," Ng said.
Ng said the TDA5 family has been launched in several markets, including Korea, and is gradually spreading. "TI's solutions are designed for deployment in the global market, including Korea," he said. "Korea is an important market for TI, and we are doing our best to support Korean customers developing next-generation Autonomous Driving technology."
"TI's core value is the passion to make the world better by providing electronic products at reasonable prices through semiconductors," he said. "That passion is fully reflected in the TDA5 family, which, despite its high technical reliability, offers low prices while reducing power consumption."