Tesla, Meta and IBM calls spotlight AI’s practical challenges

Tesla, Meta and IBM executives highlighted different challenges facing artificial intelligence during their latest earnings calls: teaching robots useful tasks, handling purchases through digital agents and demonstrating financial returns. Yahoo Finance’s account of the calls showed how commercial execution is becoming central to the discussion alongside technological capability.

Tesla reduced its annual capital spending plan to $20 billion from the $25 billion outlined three months earlier, according to Yahoo Finance. The report identified the training still needed for its Optimus robots as an apparent factor, rather than presenting that explanation as a fully established breakdown of the spending change.

Elon Musk described a fundamental difference between developing Optimus and training vehicle systems. Tesla can gather information from its cars, he said, while robots must build experience through practice and self-play. His comparison with a child learning everyday skills underscored that producing hardware and making it useful are separate challenges.

At Meta, Mark Zuckerberg described shopping as the most common use for the Muse AI agent. More than a quarter of users were engaging in shopping, with average purchases exceeding $200, according to the remarks reported by Yahoo Finance. He said activity exceeded expectations enough that Meta worked with Stripe to raise credit limits during the first week.

Those comments suggest substantial early transaction demand, but they do not establish that consumers are overspending or failing to repay debts. Yahoo Finance raised the possibility of future losses for retailers and banks as a question, not a documented outcome. Higher payment capacity alone cannot resolve that distinction.

Stripe’s September announcement provides context for how the purchasing system works. It said US consumers could connect their Link accounts to Muse, enabling checkout at more than 1 million businesses accepting Link. For other merchants, Link could issue a single-use virtual card restricted to an approved purchase, illustrating the payment infrastructure behind the agent experience.

Meta has separately emphasized authorization in its expansion of Muse for small businesses. Its announcement said the service could connect with existing business tools, while requiring approval before publishing, sending or spending. That stated safeguard is relevant to assessing agent capabilities, although a product policy is not evidence about every transaction’s outcome.

IBM chief executive Arvind Krishna described another commercial test: whether enterprise customers can justify their AI bills. According to Yahoo Finance, he said larger businesses were examining returns more closely and worrying about spending without clear benefits. He nevertheless maintained that the technology’s productivity advantages were real.

IBM’s third-quarter results announcement offered a broader financial backdrop, reporting revenue of $17.1 billion and raising its full-year free-cash-flow outlook. The company maintained its forecast for constant-currency revenue growth of 4% to 5%. Those companywide figures provide context for Krishna’s confidence, but do not independently measure the returns customers receive from individual AI projects.

Together, the disclosures point to different milestones rather than a single verdict on AI investment. Tesla’s challenge is practical robot capability; Meta’s is translating agent adoption into dependable commercial activity; IBM’s customers are weighing costs against benefits. Further disclosures will be needed to establish how consistently those ambitions translate into useful products and financial results.