Even before we go deeper, I want to be upfront about this. I think Dario Amodei’s concerns about the pace of artificial intelligence (AI) development deserve to be taken seriously.
As the CEO of Anthropic, one of the companies building this technology, he has a front-row view of where AI is heading. That alone gives his warnings a perspective most of us don’t have.
Dario Amodei published an essay titled “We Must Pace the Frontier” on Sep. 12, 2026, calling for an intentional slowdown in advanced AI development.
He received remarkable support. OpenAI CEO Sam Altman endorsed it. Elon Musk said Amodei was “right.” Even competitors in a fiercely competitive industry agreed that independent evaluators made sense.
This has been a primary topic of discussion for some time. I also think that if AI development continues at its current pace without safeguards, there is a real risk that we could eventually confront the darker side of what the technology is capable of doing.
Yes, of course, the potential benefits are enormous, but so are the consequences if its risks outpace our ability to manage them.
Hock Tan was on Mad Money on Sep. 14 and said that none of this changes anything for Broadcom (AVGO).
Both things can be true simultaneously. And understanding the tension between them is the most important thing for Broadcom investors right now.
What Broadcom CEO actually said and what he didn’t dismiss
The market’s reaction to Amodei’s essay was expected. AVGO fell over 4.8% on Sep. 14. The iShares Semiconductor ETF dropped 5.6%. Infrastructure providers across the board got hit.
And most investors’ question was whether a slowdown in frontier model development means a slowdown in compute demand.
Cramer asked Tan whether any of this has caused him to reconsider Broadcom’s fiscal 2027 and 2028 AI semiconductor forecasts.
No, not in the least.
“We see the demand for compute infrastructure, for AI development or AI frontier models, and inference for the products that they feed to the world, as continuing to be very strong and, I believe, very durable,” Tan said.
Tan wasn’t dismissing Amodei’s concerns. Tan agreed that AI needs guardrails.
“Like any tool, it’s important to put governance and safeguards on how we use the tool,” Tan said.
But he pushed back on the framing that AI poses existential risk at current capability levels. “It’s not a live animal that will run wild by itself.”
More AI Stocks:
- Wells Fargo strongly resets Marvell stock target before earnings
- Marvell’s $120 billion deal with Google has fine print
- Analyst rethinks Nvidia stock ahead of earnings
His more optimistic framing compared generative AI to the Industrial Revolution, a disruptive force that demanded major adaptation but ultimately raised living standards and expanded what humans could accomplish.
That’s a reasonable position right there. It’s also the position you’d expect from the CEO of a company whose entire growth trajectory depends on continued AI infrastructure spending.
The Broadcom numbers that make Tan’s confidence legible
Broadcom reported Q3 fiscal 2026 results that give his confidence a factual foundation.
- Q3 AI semiconductor revenue came in at $16.7 billion, up 221% year-over-year (YoY) and 54% quarter-over-quarter.
- Total consolidated revenue hit $29.6 billion, up 86% YoY, with free cash flow of $13.7 billion — 46% of revenue.
- Q4 guidance calls for AI semiconductor revenue of $21.7 billion, up 236% YoY, and consolidated revenue of $34.8 billion, up 93% YoY
- Full fiscal 2026 AI revenue is now expected to be $58 billion, raised from prior guidance of $56 billion.
The multi-year targets are even wilder. On the earnings call, Tan guided fiscal 2027 AI semiconductor revenue to approximately $115 billion, doubling again to approximately $230 billion in fiscal 2028. He reiterated confidence in exceeding $30 in EPS by fiscal 2028.
What makes these targets credible is the customer visibility behind them. Anthropic is on track to become Broadcom’s largest XPU customer in 2027-2028, with visibility to 10 gigawatts of deployment by 2028, according to the earnings call transcript.
Why inference demand is the argument Tan is really making
I also think Tan made an important distinction on Mad Money that deserves some attention. He separated training, which involves building new frontier AI models, from inference, which is the process of deploying those models and serving users at scale.
“I don’t know about training, but when you want to productize inference, I see it continuing to be very, very strong,” Tan said.
Even if Amodei’s essay succeeds in moderating the pace of training new frontier models, the inference demand from deploying existing models continues growing. Every ChatGPT query, every Claude conversation, every AI agent running in enterprise software requires compute. That compute doesn’t slow down if training slows down.
Also Read: Broadcom Inc. Latest News and Stories
Tan also helped establish a $35 billion AI special-purpose vehicle platform with Apollo and Blackstone to support Anthropic’s one-gigawatt deployment.
The commitment sits somewhat awkwardly alongside Amodei’s caution. But it also addresses clearly the point that capital is still pouring into AI infrastructure, regardless of the debate over how quickly the models themselves should advance.
Broadcom shares are down 1.44% year-to-date and 5.97% over the past year, according to Yahoo Finance. That weakness stands out against the company’s 86% YoY revenue growth and reflects the expectations gap Morgan Stanley flagged in my earlier pre-earnings report.
As mentioned before, the main problem with AVGO performance is expectations rather than fundamentals. The business is growing at a remarkable pace. The stock, however, isn’t getting the same memo.
Related: Jim Cramer has strong message for Nvidia, Broadcom investors