First things first

Earlier this week, I linked to a story reporting that chip-design firm Arm was buying big data startup Treasure Data for $600 million, and today Arm confirmed that acquisition (although not the price) and explained how Treasure will fit into a new data analytics platform called Pelion. Arm's blog post certainly answered some of the questions I had -- that Treasure Data will continue to operate as a standalone business, for starters -- but now the major question is how Arm will fare as an IoT platform provider. The company certainly is in a unique position to make a play at this burgeoning market, but competing against the big three cloud providers, all of which are stepping up their IoT stories, is always an uphill battle.

Here are two other items from today that warrant a little more attention:

  • A new era in data warehousing (Cloudera): When I used to cover the Hadoop space more closely, Cloudera would often claim that it didn't want to take on data warehouse vendors directly. Well, things have obviously changed with this new product (or maybe just a new name on an existing product?). However, it seems likely the difficult part for Cloudera won't just be winning deals away from existing data warehouse vendors, but rather winning them away from cloud-provider services and cloud-native (in a different sense of the word) data warehouse startups such as Snowflake and the recently launched Yellowbrick.
  • Tesla is building its own AI chips for self-driving cars (TechCrunch): This makes perfect sense for all the reasons laid out in the post, but a couple of questions do come to mind. One is the pace at which Tesla will have to evolve its chips, or build new ones, to handle new tasks or improve performance/efficiency. Over time, the costs might outweigh the immediate performance hit (to the degree it's material) of using an industry-standard platform. The other question is what happens to independent auto mechanics as cars get more advanced and more customized. It could be all but impossible to learn the specifics of each maker's AI systems, but on the other hand software fixes could be fixed over the air and hardware fixes might just requires a few screws.
Read and share this issue online here.

AI and machine learning




Cloud and infrastructure






Data and analytics