More than three dozen venture capitalists, chip designers, and top-tier tech executives packed a private room at San Francisco’s InterContinental Hotel last Tuesday, and the chatter wasn’t about NVIDIA’s latest GPU or TSMC’s 2nm roadmap. It was all about Intel—specifically, the audacious, customer-first foundry plan that new CEO Lip-Bu Tan laid out over coffee and canapés. By the time the two-hour session wrapped, several attendees had reportedly pulled out their phones to snap photos of Tan’s slide deck, a gesture that in Silicon Valley signals genuine intrigue rather than mere politeness.
That moment, described by three people familiar with the event, captures the mood shift that has lifted Intel’s stock 18% in the past month and added roughly $35 billion to its market capitalization. After years of manufacturing missteps, market-share losses, and a pervasive sense that the company had missed the AI revolution, the semiconductor giant is suddenly back in the conversation—and this time, the believers aren’t just long-suffering retail investors.
From Gelsinger’s IDM 2.0 to Tan’s Collaborative Pivot
Intel’s turnaround story has been told many times before. Pat Gelsinger’s return as CEO in 2021 brought with it a sweeping “IDM 2.0” strategy that aimed to restore process leadership by 2025 and build a world-class foundry business. Yet despite record spending on new fabs in Ohio and Germany, Intel’s execution faltered. Delays in the 10nm and 7nm nodes cost it the lead to TSMC, while AMD captured 30% of the PC CPU market for the first time in over a decade. Gelsinger’s departure in late 2023 left a leadership vacuum that the board struggled to fill.
Enter Lip-Bu Tan. A former Cadence Design Systems CEO with deep roots in the fabless ecosystem, Tan joined Intel’s board in March 2024 and was named interim chief executive that December. By mid-2025, the “interim” tag had been removed, and he immediately set about reshaping the company’s narrative. His core thesis: Intel’s manufacturing prowess, long treated as a proprietary weapon, must become an open, collaborative service that puts customer needs ahead of Intel’s own product timelines.
Inside the San Francisco Summit
The San Francisco gathering, officially titled “Foundry Forward: Intel’s Next Chapter,” wasn’t an investor conference but an invitation-only dialogue with chip designers and potential foundry customers. Attendees included executives from Qualcomm, Broadcom, and several AI startups, all of whom currently rely on TSMC for advanced nodes. Tan, speaking without a teleprompter, walked through a revamped process roadmap that highlights Intel’s 18A node—hitting its “best-ever” defect density metrics in Q2 2025—and previewed a “customer-first” engagement model that promises capacity guarantees and design support previously unheard of at Intel.
“He didn’t sugarcoat the past mistakes,” one attendee said. “He said, ‘We lost your trust. Now we’re going to earn it back one tape-out at a time.’” That directness, combined with tangible process improvements, created what one longtime industry analyst calls the “belief phase” of Intel’s comeback. In semiconductor turnarounds, this phase is critical: it’s when theoretical roadmaps start convincing partners to commit real design resources.
A Customer-First Foundry in Practice
What does a “customer-first” Intel foundry actually mean? Historically, Intel’s fabs were optimized for its own CPUs. Third-party customers were often relegated to lower-priority production slots, and the company’s design tools were notoriously rigid. Tan is dismantling that model. Under his restructuring, Intel Foundry Services (IFS) operates as a semi-autonomous business unit with its own P&L, sales force, and—crucially—a streamlined interface that supports industry-standard EDA tools from Cadence, Synopsys, and Siemens.
A key differentiator is what Tan calls “Open Cell Architecture.” Instead of locking customers into Intel’s standard cell libraries, IFS now allows chip designers to bring their own optimized cells, then pairs them with Intel’s advanced packaging technologies like EMIB and Foveros Direct. This approach makes Intel a viable alternative to TSMC for the first time in a decade, particularly for companies designing high-performance AI accelerators that need both leading-edge logic and heterogeneous integration.
The AI Chip Imperative
Nowhere is this shift more critical than in the AI chip market. NVIDIA’s dominance has left the entire industry scrambling for alternatives, and hyperscalers like Microsoft, Amazon, and Google are increasingly designing their own silicon. Intel’s new foundry strategy positions it as a neutral manufacturing partner that can fabricate custom AI chips without competing on the design side—a sharp contrast to NVIDIA’s closed ecosystem.
Tan has already signed a multi-year agreement with a major cloud provider (rumored to be Microsoft) to produce custom inference chips on the 18A node. The deal, first hinted at during Intel’s Q2 earnings call, underscores a broader re-alignment: Windows PC makers, too, are watching closely. Qualcomm’s Snapdragon X Elite has proven that Arm-based SoCs can beat x86 in battery life and AI performance, and Intel must respond not only with better CPUs but with a foundry model that could actually fabricate Qualcomm’s chips if the partnership evolves.
Windows PC Ecosystem at a Crossroads
For the Windows ecosystem, Intel’s revitalization is existential. More than 70% of all Windows PCs still ship with Intel processors, but that share is under assault from multiple directions. AMD’s Ryzen AI 300 series now offers neural processing unit (NPU) performance that exceeds Intel Meteor Lake’s 10 TOPS, and Apple’s M-series chips continuously demonstrate what a vertically integrated design- manufacturing loop can achieve.
Tan’s foundry vision could redefine the PC supply chain. If Intel’s fabs become the go-to source for premium notebook SoCs—whether they carry an Intel, AMD, or even a Qualcomm badge—Windows laptops would benefit from a broader silicon pipeline and faster iteration cycles. OEMs like Dell and Lenovo, bruised by recent chip shortages, have privately expressed support for any move that diversifies their advanced-node sourcing.
“We design our machines around the chip availability,” a senior engineering manager at a major PC OEM told windowsnews.ai. “When Intel said they’d guarantee 18A wafers for external customers, our ears perked up. A healthy Intel foundry means a healthier Windows PC market.”
Market Reaction and the $35 Billion Signal
Wall Street rarely buys into semiconductor comeback stories without concrete proof, but the numbers are starting to align. Intel shares have surged from the mid-$20s to nearly $40 in the past six months, with the San Francisco summit sparking the latest leg higher. Analysts at Morgan Stanley and UBS have upgraded Intel’s rating, citing “improved foundry traction” and “credible management vision.” The $35 billion market-cap gain since Tan’s permanent appointment isn’t just a relief rally; it’s a bet that Intel can split the foundry duopoly now held by TSMC and Samsung.
Yet skepticism persists. Intel’s capital expenditure remains a staggering $25 billion per year, and its cumulative net debt has ballooned to $52 billion. For the customer-first model to work, IFS must generate at least $15 billion in external revenue by 2028—a target that assumes several major customers migrate their leading-edge designs away from TSMC. The Microsoft deal is a beachhead, but converting it into a portfolio of blue-chip clients will require flawless execution.
The Road Ahead: Windows 12 and the AI PC Wave
Microsoft’s impending Windows 12 launch, expected in early 2026, will place AI capabilities at the center of the user experience. Copilot+ features—local language models, real-time transcription, AI-enhanced search—demand NPUs with at least 40 TOPS. Intel’s Panther Lake architecture, slated for late 2025, promises a tripling of AI TOPS over Meteor Lake, but only if Intel’s fabs can deliver on schedule.
Here, Tan’s foundry-first mentality creates a virtuous circle. By treating IFS as a separate entity, Intel can prioritize Panther Lake’s ramp without cannibalizing capacity intended for external customers. This dual-track approach contrasts with Gelsinger’s era, when internal product groups often competed with foundry operations for the same wafers. The result: Windows users may see a wider array of NPU-rich chips hitting the market simultaneously, from Intel’s own Core Ultra line to AMD’s Zen 6-based APUs and even Qualcomm’s next-generation Oryon cores—all potentially built on Intel nodes.
Challenges That Could Derail the Belief
Even Tan’s most enthusiastic supporters acknowledge that Intel’s path remains littered with obstacles. The 18A node, while promising, must scale to high-volume manufacturing without a repeat of the 10nm debacle. Geopolitics add another layer: Intel’s new fabs in Arizona and Ohio rely on CHIPS Act subsidies that could be revised under a new Congress. And talent retention is a constant battle, with Apple, NVIDIA, and AMD poaching top engineers from Hillsboro and Santa Clara.
There’s also the cultural question. Intel’s historical arrogance—a “not invented here” syndrome—runs deep. Tan has moved quickly to replace key lieutenants and bring in outsiders, but changing the DNA of a 120,000-employee organization takes years, not months. The belief phase can curdle into disappointment if the first external tape-outs face unexpected yield issues or poor support.
What Windows Enthusiasts Should Watch
For those who follow the Windows hardware beat, several landmarks will signal whether the comeback is real: first, the Q3 2025 earnings call, where Tan is expected to disclose IFS revenue breakout and the number of active design engagements; second, the rumored Microsoft custom-chip prototype, which could appear in a Surface device by mid-2026; and third, the public launch of Intel’s 18A-powered Panther Lake chips at CES 2026.
Pay attention to the language OEMs use during laptop launches. If partners like ASUS and Acer begin mentioning “multi-sourced silicon” or “foundry flexibility,” it’s a tip that Tan’s customer-first message is resonating beyond the InterContinental ballroom. In the meantime, the Windows PC market is getting something it hasn’t enjoyed in a long while: genuine competitive tension at the most critical layer of the stack.