This story is available exclusively to Business Insider subscribers. Become an Insider and start reading now. Tesla appears to be getting closer to deploying Optimus in its factories — and Austin is the next training ground. Tesla told workers during a town hall last week that it plans to begin collecting data to train its humanoid robot at its Austin Gigafactory, insiders told Business Insider. T...
This story is available exclusively to Business Insider subscribers. Become an Insider and start reading now. Tesla appears to be getting closer to deploying Optimus in its factories — and Austin is the next training ground. Tesla told workers during a town hall last week that it plans to begin collecting data to train its humanoid robot at its Austin Gigafactory, insiders told Business Insider. The company is looking to train Optimus how to operate in the Texas facility, it said, adding that it was targeting a February start date. Tesla has been collecting data and training Optimus prototypes in its Fremont, California, factory for more than a year. The Optimus data collectors in Fremont are typically kept separate from the general factory workers to avoid interfering with output, people with knowledge of the program said. There, data collectors have recorded themselves organizing vehicle parts and working on conveyor belts, Business Insider previously reported. The videos are then used to teach Optimus how to mimic the same movements. CEO Elon Musk said during an interview at Davos on Thursday that Tesla has the humanoid robot doing "simple tasks" in its factory, though he didn't specify which. "By the end of this year I think they'll be doing more complex tasks but still deployed in an industrial environment," he said. "By the end of next year I think we'll be selling humanoid robots to the public." He warned earlier this week in a post on X that production of Optimus — and the company's Cybercab, which will be produced in Austin — will be "agonizingly slow." A spokesperson for Tesla did not respond to a request for comment. Tesla has been sharing sneak peeks of Optimus' capabilities since the product was first announced in 2021. In 2024, the automaker posted a video that showed the robot arranging batteries while tied to an overhead support structure at the company's lab in Palo Alto. The company said that it deployed two autonomous Optimus robots in one of its ...
Key PointsThe Dow Jones Industrial Average, S&P 500, and Nasdaq Composite have soared to new heights on the heels of the AI revolution and the Federal Reserve's rate-easing cycle.
Key PointsThe Dow Jones Industrial Average, S&P 500, and Nasdaq Composite have soared to new heights on the heels of the AI revolution and the Federal Reserve's rate-easing cycle.
Micron Technology's advanced memory chips are proving essential in the AI race. When you think of artificial intelligence (AI) semiconductors, you tend to think about the graphics processing units (GPUs) from companies like Nvidia (NVDA +1.60%) that perform the actual computations that AI requires. But not all semiconductor chips are the same, and not all of them are made by Nvidia. In fact, some ...
Micron Technology's advanced memory chips are proving essential in the AI race. When you think of artificial intelligence (AI) semiconductors, you tend to think about the graphics processing units (GPUs) from companies like Nvidia (NVDA +1.60%) that perform the actual computations that AI requires. But not all semiconductor chips are the same, and not all of them are made by Nvidia. In fact, some of the most important resources for Nvidia's GPUs are the memory chips that keep data accessible. And the memory chip company that's quietly becoming a cornerstone of the AI boom is Micron Technology (MU +0.52%). Not all memory is the same Old-school computer users may remember the two basic memory types: Random access memory (RAM) stores data that's currently being used by the processor, while read-only memory (ROM) is for more permanent storage. But within these categories, there are plenty of subtypes. One important RAM subtype is dynamic random access memory (DRAM). A DRAM chip consists of a series of simple memory cells, each containing just one capacitor. A charged capacitor indicates a one, and a discharged capacitor indicates a zero. Because each memory cell is so simple, DRAM is cheap to produce and can hold large amounts of memory in a single chip. However, capacitors leak charge quickly, so DRAM needs to be recharged every few milliseconds, making it very energy-hungry. AI computations need fast access to more data than a single DRAM wafer can allow due to their massive data needs and high processor speeds. Luckily, by stacking DRAM wafers vertically and adding some extra electrical connections, you can create high bandwidth memory (HBM), which delivers incredible data retrieval speed for less power. Only three companies currently manufacture almost the entire global supply of DRAM and HBM: SK Hynix and Samsung, both from South Korea, and Micron. Expand NASDAQ : MU Micron Technology Today's Change ( 0.52 %) $ 2.08 Current Price $ 399.66 Key Data Points Market Cap...