Intel CEO gives investors a reality check

AI stock investors have essentially been trained to follow a rule to invest in businesses closest to GPUs, data centers, and hyperscaler spending.

Intel (INTC) CEO Lip-Bu Tan just flipped that script on the June 18 episode of No Priors: AI, Machine Learning, Tech, & Startups. 

Tan believes that the AI trade is effectively moving beyond a simple GPU story and into harder infrastructure layers behind it 

He effectively laid out a far bigger map of where the semiconductor space is being stretched. 

That gives Intel a remarkably ambitious investor narrative, but also a tougher test in proving it can execute where the entire AI supply chain is under pressure.

Intel CEO points to AI’s next chip bottlenecks 

Intel CEO Lip-Bu Tan used the No Priors podcast to push the AI investment debate beyond GPUs.

More AI:

Perhaps his biggest rationale was that agentic AI is changing the compute mix. 

Tan argued that the old training setup relied too heavily on GPUs, but newer AI workloads require much more orchestration, reinforcement learning, and coordination across agents. That is bringing CPUs back into the discussion. “Right now, the demand is very high for my CPU,” he said.

For perspective, the AI trade has been dominated by demand for accelerators.

However, Tan feels the next phase also depends on the infrastructure around the accelerator: CPUs, memory, interconnects, packaging, power, and manufacturing capacity.

He also flagged memory as a pressure point, saying “memory is a bigger shortage” as companies scramble for supply. Additionally, power is another constraint. Tan said some countries simply do not have enough power capacity to support AI growth, while power conversion and thermal limits are becoming more important across the chip stack.

Moreover, Tan said a foundry is a “service business” and a “trust business”, where customers care about yield, defect density, cycle time, and reliability. If a chipmaker misses the mark on those points, the customer can lose sales.

Hence, AI demand is far from being just a one-stock or one-chip story. Tan is pointing to a broader semiconductor bottleneck cycle, in which the winners are likely to be those that solve the physical limits constraining AI growth.

Big-bank Intel stock targets show Wall Street is split after the AI rally

Bank of America: $135. BofA double-upgraded Intel to Buy from Underperform and lifted its target from $96, citing stronger AI server CPU demand, foundry customer momentum and higher 2030 earnings power.

Citi: $130. Citi raised its Intel target from $95, arguing that agentic AI could sharply expand the server CPU market and give Intel a bigger data-center growth runway.

Mizuho: $135. Mizuho raised its target to $128 while maintaining a Neutral rating, citing a better outlook for advanced packaging, foundry platforms, and domestic chip manufacturing demand.

Wells Fargo: $110. Wells Fargo lifted its target from $85 but maintained an Equal Weight rating, suggesting Intel’s AI CPU and foundry story is improving, but the stock already reflects much of that optimism.

Barclays: $100. Barclays raised its target from $65 while keeping an Equal Weight view, signaling more confidence in Intel’s turnaround, but continued caution after the stock’s huge run.Sources: Yahoo Finance, TheStreet, Investopedia, Trading View

Related: Lip-Bu Tan’s net worth: The Intel CEO’s wealth & income

What Intel still has to prove in foundry and what comes next for AI stock investors

Intel’s foundry story has clearly evolved over the past year or so.

The latest figures show why the market remains cautious. 

In Q1 2026, Intel Foundry reported $5.4 billion of segment sales, up 16%, but intersegment eliminations were $5.3 billion, meaning the business still largely supports Intel’s own products. The segment also posted a $2.4 billion operating loss, while Intel Products earned $4.1 billion.

The full-year picture is similar. 

Intel Foundry generated $17.8 billion of 2025 revenue, but only $307 million came from external customers. Its operating loss was $10.3 billion.

That is why the next proof point is all about real outside customer orders, produced at yields strong enough to support volume manufacturing.

There has been some strong progress, though, of late.

Intel said 18A-P entered risk production, with 9% better performance at the same power, 18% lower power at the same performance, and 20% to 40% better thermal resistance.

Intel also brought in Seok-Hee Lee to lead advanced packaging and back-end manufacturing, areas that matter more as AI chips become multi-die systems.

For AI stock investors, the takeaway goes beyond Intel.

Tan’s message points to a market moving from GPU scarcity to supply-chain scarcity. CPUs, memory, packaging, power, and foundry capacity are becoming investable bottlenecks. Naturally, Intel can benefit from that shift, but only if it proves customers can trust its factories at scale.

Related: Tesla rival says Elon Musk still has one major edge

This story was originally published by TheStreet on Jun 22, 2026, where it first appeared in the Investing section. Add TheStreet as a Preferred Source by clicking here.