|
Amazon prides itself on being data obsessed, and its approach to tracking its own employees’ use of AI has been no different. Amazon has been using an internal system called Clarity to track how often employees use AI tools in their work, two people with knowledge of the tracking said. (Clarity also tracks other things about employees, such as in-person office attendance, the people said.) That includes data on overall AI usage by team and—for at least some managers—details on which AI tools people are using, including Amazon’s coding tool, Kiro. Amazon says it encourages employees to use its in-house AI tools as well as certain approved tools made by other companies. "Understanding how employees adopt new technology helps us support them in using the latest tools to innovate in their day to day and deliver for customers,” an Amazon spokesperson said. “We focus on AI adoption and sharing best practices to celebrate innovation and operational efficiency gains across the company—whether that’s during a review process or throughout the year.” Apart from tracking time spent on AI tools, some leaders in Amazon have been asking for more details on how employees are using AI to work more efficiently. That includes the company’s supply chain optimization technology team, known as SCOT, which has played a key role in managing costs in Amazon’s massive retail and logistics business, such as figuring out how to balance speedy delivery with keeping fulfillment costs low. In that unit, assessments for all employees up for potential promotion now include questions about how they are using AI. Previously, questions on AI usage had only been included in assessment for promotions above L7, a middle management role. In a July memo announcing the change, Matt Taddy, vice president of SCOT, said it was made in part to align promotions with “impact, efficiency, execution and not org size.” For instance, a new question for SCOT employees asks for a description of how that person used AI to “innovate, improve customer experiences, or enhance operational efficiency or effectiveness in their role or for their team,” including specific examples and measurable outcomes. Amazon is among a growing list of tech giants that are tracking workers’ internal AI usage and using it as a factor in decisions on pay and promotions. In January, Meta told employees that performance reviews and bonuses would weigh metrics including generating code with AI tools. And Accenture recently started collecting data on employee AI logins and told staff that promotions to senior levels would require “regular adoption” of the technology. In some cases, tech companies are encouraging employees to use the company’s in-house tools. Microsoft has told enterprise clients it will be “customer zero” for the AI tools it is selling, while telling some employees to increase their use of its AI coding tools to generate more of the code they write. These kinds of tracking and performance management decisions can ruffle employees’ feathers, given that the changes sometimes highlight how well staffers are automating work currently done by humans. And in Amazon’s case, some of the efforts linking AI usage to promotions have come as the company is cutting corporate jobs, including 14,000 in October and another 16,000 in January—cuts the company said will reduce bureaucracy and make it operate more nimbly. Amazon has said recent job cuts weren’t directly AI driven, with CEO Andy Jassy saying in an interview with The Information in January that “we announced some role reductions, people assumed it was just AI, because AI is kind of in the ethos of everything right now. And they just weren’t AI. They were really about our culture.” The company also continues to hire in high-priority areas. Still, promotion assessments for people in Amazon’s SCOT unit who are already managers also ask how they have “accomplished more with less” and “specific examples of how they have remained innovative, force multiplied using AI and delivered results while reducing or not growing headcount,” according to the memo. One corner of enterprise software is proving less vulnerable to the fears of potential AI disruption that have hammered the stocks of some companies over the past several months. Cadence Design Systems’ shares rose 7.6% Wednesday after its fourth-quarter earnings report the previous day showed its core software business continuing to grow. The company is one of two leading chip design and verification software firms, the other being Nvidia-backed Synopsys. Both companies sell tools used by chipmakers, including Nvidia, Broadcom, Qualcomm, Altera and Tenstorrent, to model their chip designs before they go to manufacturers such as Taiwan Semiconductor Manufacturing Co., Intel or Samsung. The goal is to make sure the chips will actually work once they’re made. As of Wednesday's close, Cadence’s stock was down just over 2% in the year to date, while Synopsys has fallen about 5.8%—somewhat of a soft punch when you consider that software industry firms in the S&P 500 have collectively tumbled about 20% in that time frame. Notably, Nvidia CEO Jensen Huang mentioned the two companies in his recent defense of software as a service at the Cisco AI Summit, when he argued that AI agents will strengthen demand for software rather than replace it. Cadence is well positioned to withstand the threat posed by AI. Unlike other software companies that have traditionally charged on seat-based pricing, it charges its customers based on their workloads. (As AI is expected to reduce the number of employees at companies, a seat-based pricing model could hurt software businesses.) Cadence also has developed agentic AI tools that autonomously generate code and run simulations for chip designs, expanding how many tasks a given engineer can handle at once. Because AI has driven an explosion in the complexity of semiconductor designs faster than chipmakers can scale up their workforces to meet ever-rising engineering demands, Cadence argues agentic AI offerings will expand rather than eat into its business. Cadence’s exposure to the semiconductor space is also a boon, as chip companies account for more than half its revenues. Chip sales are, of course, booming thanks to AI. More chip designers are entering the playing field, and Cadence executives said Wednesday that cloud players’ increasing bets on in-house silicon will boost its business. And finally, there’s the fact that Cadence and its rival Synopsys are simply too complex for new AI tools to easily displace, KeyBanc Capital Markets analyst Jason Celino said. As he put it, “Designing a semiconductor chip is literally one of the most difficult things to do mathematically.…The only harder thing than designing a semiconductor chip is rocket science.”—Laura Bratton
|
