Intel Corp Intel (INTC) shares were trending lower in Tuesday's premarket, down about 1.95% to $107.20, as the broader market took a breather. But beneath the surface, the company is making some big moves at Computex 2026, trying to convince investors that its AI story is about more than just catching up to NVIDIA.

At the conference, Intel laid out its "chip-to-rackscale" AI vision, emphasizing inference workloads—the part of AI where models actually respond to user requests—rather than the training-heavy approach that has made NVIDIA the undisputed king. The idea is that as AI shifts toward agentic applications (think AI agents that can book your flights or manage your calendar), the data center will need more CPUs relative to GPUs. Intel is betting that this shift plays to its strengths.

Creative Strategies CEO Ben Bajarin has argued that agentic inference could push data centers toward a one-CPU-to-one-GPU (or less) ratio, compared to the training-era pattern of one CPU per four GPUs. If that plays out, Intel's Xeon processors could become a much bigger part of the AI infrastructure story.

Crescent Island: Intel's Inference Gambit

Intel's new AI-focused GPU, Crescent Island, is at the heart of this strategy. The company plans to begin limited shipments by the end of 2026. Kevork Kechichian, who leads Intel's data-center group, told the Financial Times that the chip is designed specifically for inference, not training. That's a deliberate choice: instead of trying to beat NVIDIA at its own game, Intel is targeting customers who care more about cost than raw training performance.

Crescent Island uses lower-cost LPDDR5 memory and air-cooling, avoiding the expensive high-bandwidth memory and liquid-cooling systems that NVIDIA and AMD rely on. That could make it a compelling option for companies running inference at scale. Kechichian also noted that Intel is evaluating whether certain versions of the chip could be sold in China while complying with U.S. export controls, citing continued demand for AI products at that price point.

Another key part of the plan: Intel wants to manufacture Crescent Island in its own foundries, part of CEO Lip-Bu Tan's turnaround strategy. That could lower costs compared to rivals that rely on Taiwan Semiconductor Manufacturing Co. Ltd TSMC (TSM).

NVIDIA Strikes Back—in PCs

While Intel is trying to claw its way back into the AI conversation, NVIDIA is opening a new front in Intel's backyard: the PC processor market. NVIDIA recently introduced its RTX Spark Superchip, developed alongside MediaTek and integrated with Microsoft Corp.'s Microsoft (MSFT) Windows platform. NVIDIA CEO Jensen Huang said the chip is designed to power AI agents that can perform tasks across multiple applications with minimal user involvement.

That's a direct challenge to Intel and Advanced Micro Devices Inc. AMD (AMD), who have long dominated the PC processor market. If NVIDIA can make inroads there, it could put additional pressure on Intel's core business just as it's trying to stabilize.

Earnings and Analyst Outlook

The next big catalyst for Intel is its Q2 2026 earnings report, expected around July 23. Analysts are looking for EPS of 19 cents (up from a loss of 10 cents a year ago) and revenue of $14.40 billion (up from $12.86 billion). The stock carries a consensus Hold rating with an average price target of $80.31—well below the current price, which suggests analysts see limited upside from here.

That said, several firms have recently raised their forecasts. On June 1, Barclays raised its target to $100, Wells Fargo to $110, and Mizuho to $128. All three maintain neutral or equal-weight ratings, so the upgrades are more about acknowledging the stock's momentum than a fundamental change in outlook.

Technical Picture: Consolidation After a Run

Intel's stock has had a massive run over the past year, more than doubling from its 200-day SMA of $49.60. But the near-term trend has cooled: the stock is trading about 7% below its 20-day SMA of $115.65, signaling a short-term pullback. The 50-day SMA sits at $83.71, meaning the stock is still 28.5% above that level—a sign that the longer-term uptrend remains intact.

The Relative Strength Index (RSI) is at 53.52, which is neutral territory. That suggests the stock isn't overbought or oversold, just consolidating. The golden cross from August 2025 (when the 50-day SMA crossed above the 200-day SMA) still supports the bullish backdrop.

Key support is at $102.50, a level where buyers have stepped in before. If Intel can hold that line, the pullback could be a healthy reset within an uptrend. If it breaks, the next support would be around the 50-day SMA.

Intel set its 52-week high of $132.75 in May, so the current move looks like a retracement from that peak. Traders will be watching to see if the stock can reclaim its 20-day SMA and build a base for the next leg higher.

For now, Intel is a story of two narratives: a company trying to reinvent itself in AI while defending its core markets, and a stock that's priced for a turnaround but faces stiff competition from NVIDIA on multiple fronts.