Jason Child on AI reliability
From EP. 11 | How Arm's CFO Sees the Future of AI and Infrastructure | Human Capital Gains · · Human Capital Gains Podcast
“I look at AI and say it's mostly a probab probabilistic uh model versus deterministic and that is it's going to give you the highest likely answer based on whatever your prompt is but it still could be wrong. And that's where we of course know all about hallucinations. Now hallucinations are significantly reducing but they still exist. And so if you're in an area like finance where we are, you know, we get a pass fail for providing determin deterministic answers to questions. Are the financials accurate? Um is the revenue correct? Uh does the revenue related to a certain deal based on the terms? Is it exactly right? Does it follow ASC 606? And you know like right now AI isn't quite ready to give you that 100% certainty. It'll give you a 90some percent certainty, but that's not enough.”
On , Jason Child, Executive VP & CFO at Arm Holdings, spoke about AI reliability during EP. 11 | How Arm's CFO Sees the Future of AI and Infrastructure | Human Capital Gains on Human Capital Gains Podcast.
About Jason Child
Jason Child, executive vice president and CFO of Arm Holdings, discussed the company's position in the AI infrastructure market during a May 2026 podcast interview. He described Arm as providing "the CPU of choice" for hyperscalers and core infrastructure providershol, and noted that the company is preparing its business to serve customers amid what he called a "significant infrastructure shift." Child distinguished current AI from earlier machine learning, stating that past recommendation algorithms were "not really intelligence" but that newer models are "getting really close to actually having some true intelligence." He advised companies to designate a person or team as an expert funnel for enterprise-grade AI solutions rather than allowing broad experimentation. On Arm's Q4 fiscal 2026 earnings call in May 2026, Child reported that data center royalty revenue "continues to more than double year-on-year," driven by arm-based server chips from major hyperscalers and data center networking chips where Arm has "close to 100% market share." He stated that the company now sees "more than 2 billion dollars of customer demand" for its Arm AE GI CPU across fiscal 2027 and 2028, double the amount cited at launch, and reiterated a target of $15 billion in revenue from that business by fiscal 2031. Child also projected total revenue of $25 billion by fiscal 2031, comprising $15 billion from the AGI CPU business and $10 billion from IP revenue, translating to "more than $9 in EPS."