Xpeng's robotics unit raises 900 million dollars as SageMaker writes 25 records per call
Xpeng's robotics unit closed a round of more than 900 million dollars at a 6.3 billion dollar valuation, which TechCrunch describes as the largest single private financing round in China's embodied AI sector. In the same period AWS added BatchWriteRecord, which writes up to 25 records in one call, and ListRecords, which paginates record identifiers, to Amazon SageMaker Feature Store. Embodied AI investment and cloud machine-learning infrastructure both grew at different scales in the same week.
Artificial Intelligence··Morning
Xpeng's robotics unit closed a record round
Xpeng's robotics unit has closed a round of more than 900 million dollars at a 6.3 billion dollar valuation, which TechCrunch describes as the largest single private financing round in China's embodied AI sector. IDG Capital led the round, with Gaorong Ventures, Tencent and Alibaba taking part. Founders He Xiaopeng and Brian Gu put in about 100 million dollars of their own money. Xpeng is developing the Iron humanoid; Chery's AiMOGA unit is preparing an IPO and BYD has unveiled a humanoid called Xiao Di. Changan, GAC, Li Auto, SAIC and Seres are working in the same area.[1]
Chinese automakers turn toward robots
Michael Dunne, who runs Dunne Insights, says profit in cars looks razor-thin in the near term while robots look more promising. TechCrunch reports that Chinese automakers are following Tesla's bet that robots are the next big profit machine. The Xpeng round stands out as the largest single private financing example of that trend.[1]
SageMaker Feature Store gained two new APIs
AWS has added two APIs to Amazon SageMaker Feature Store. BatchWriteRecord writes up to 25 records across several feature groups in a single call, with the write succeeding or failing per record and a per-record time to live. ListRecords enumerates record identifiers in a feature group with pagination and works on both the DynamoDB-backed standard tier and the Redis-backed in-memory tier. PutRecord needed a separate call per record: in the example AWS gives, a fraud-detection pipeline pushing 10,000 records a second across 5 feature groups has to sustain 50,000 calls a second. ListRecords returns 10 records a page by default and 100 at most.[2]
