Wall Street analysts have turned bullish on Taiwan Semiconductor Manufacturing Company (NYSE: TSM), arguing that the rise of agentic AI (software agents that plan and act autonomously, rather than just answer prompts) will push the foundry to double down on capital expenditure, Benzinga reports.

The report frames the call simply. Demand for AI compute keeps climbing, and TSMC manufactures the advanced chips that demand runs through. Agentic workloads multiply inference (running a trained model to get outputs) because agents call models repeatedly as they work through multi-step tasks. More inference means more silicon. More silicon means more fabs, more tools, more spend.

TSMC did not announce new spending figures in the report. The bullish case rests on analyst expectations that the company will keep expanding its already heavy investment program rather than hold it flat.

What does this change for investors?

For holders of TSM stock, the analyst view lands on one question: is rising capital expenditure a burden or a moat? The bulls' answer, as summarized by Benzinga, is the second. Heavy spending on leading-edge capacity is what lets TSMC capture AI demand that competitors can't serve. Spending now is positioning, not waste.

The risk side is the usual one. Capital expenditure has to be paid for before revenue arrives, and AI demand forecasts have already been revised hard in both directions over the past two years. Agentic AI is young; nobody has settled numbers on how much inference volume it will actually generate.

Why agentic AI matters to a chip factory

The shift from chatbots to agents changes usage patterns, not just capabilities. A single agent task can trigger dozens of model calls, tool uses and memory reads. Each call burns compute on GPUs and accelerators, most of which are fabricated by TSMC on its leading-edge process nodes (the most advanced chip manufacturing tiers a foundry offers).

That volume logic is the core of the analyst thesis reported by Benzinga: agentic AI expands the total amount of inference the industry runs, and expanding inference expands TSMC's addressable demand.

The report does not name the specific analysts behind the calls or attach price targets. It characterizes the sentiment as bullish on the stock alongside the spending outlook.

The backdrop

TSMC remains the dominant contract chipmaker, producing advanced processors for the major AI accelerator vendors. Its capital expenditure is among the largest of any company in the semiconductor supply chain, and its capacity decisions effectively set the pace for how fast global AI compute can scale.

Investors tracking the story will find the next hard data points in the company's upcoming earnings release and capital expenditure guidance, where analysts' expectations either get confirmed or get repriced.

By Grace Kim