Intel CEO Patrick P. Gelsinger stated that his target for Intel is a "10x return within 5 to 10 years," and he is systematically redefining Intel’s technology roadmap around advanced packaging, novel semiconductor materials, and next-generation substrate technologies.

In a recent podcast appearance, Gelsinger elaborated on his transformation strategy: after stabilizing the balance sheet and focusing on product lines, he is now shifting strategic emphasis toward advanced packaging technologies such as EMIB, glass substrates, and emerging materials including gallium nitride (GaN), silicon carbide (SiC), indium phosphide (InP), and synthetic diamond. These efforts aim to address the growing physical limits of traditional process node scaling. He also revealed that the surge in AI agents and inference workloads is driving a strong rebound in CPU demand, with the CPU-to-GPU ratio in data center servers evolving from past ratios of 1:8 toward 1:4 or even lower.

Gelsinger noted that Intel has already delivered approximately a 6x return to shareholders over the past 14 months—but “this is just the beginning.” He expects that by 2030 to 2032, the market will begin to truly recognize Intel’s potential—not limited to its traditional PC client base, but extending into emerging domains such as edge computing, physical AI, and agent-based AI.

To him, if Intel’s XPU architecture, advanced packaging capabilities, and foundry strength are effectively integrated, they can deliver customized chip solutions tailored to diverse workloads—this is the long-term strategic direction he has anchored for the company.

New Materials as the Key to Breakthrough: Advanced Packaging and Glass Substrates at the Core

As traditional process node scaling increasingly approaches physical limits, Gelsinger has identified materials science and advanced packaging as the primary breakthrough vectors. He noted that Intel has already mass-produced its 18A process, is advancing toward 14A production, and can see viable paths to 10nm and even 7nm—but “this path will become increasingly expensive and difficult.”

To counter this, Gelsinger has launched multiple initiatives in packaging materials. He invested in 3DGS, a glass substrate company, drawn by glass’s unique thermal insulation and dielectric properties. In inter-chip interconnects, Intel is pushing forward its next-generation advanced packaging technology—EMIB—and has announced new collaborations for advanced packaging manufacturing in India and New Mexico, USA. With around 1,000 patents in module technology, integrating substrates and modules effectively remains a core engineering challenge emphasized by Gelsinger.

In semiconductor materials, Gelsinger confirmed investments in GaN, SiC, and InP, with some of these ventures already acquired by major players like Analog Devices (ADI). He also invested in a synthetic diamond wafer company, recognizing diamond’s potential as a high-performance thermal management material in chip packaging. “That’s the engineer’s mindset—you keep hitting bottlenecks, then you figure out how to overcome or bypass them,” he said.

Foundry Business: Trust First, Yield and Cycle Time Are Key Metrics

Intel’s foundry business was once considered unsustainable by many external observers, but Gelsinger chose to stay committed. His core rationale: domestic advanced manufacturing in the U.S. holds strategic value for supply chain security. No major semiconductor company should concentrate its supply chain in just one or two geographic regions.

Operationally, Gelsinger has prioritized yield, defect density, and cycle time as key performance indicators for the foundry business. He stressed that foundry is fundamentally a trust-driven business: “Customers must trust you before they hand over their wafers.” Once yield fails to meet targets, revenue losses drive customer attrition—making it nearly impossible to recover.

He also clarified that Intel and TSMC maintain a partnership relationship, not merely competitive rivalry. The industry as a whole needs more capacity to meet rising demand. He anticipates that Intel’s foundry business will begin demonstrating its true potential in the market by 2030 to 2032.

Terafab Collaboration: Building Semiconductor Infrastructure with Elon Musk

Gelsinger revealed that the Terafab project with Elon Musk stems from a shared conviction: semiconductor infrastructure development lags behind the explosive growth of AI demand in terms of capacity, production efficiency, and power efficiency. Under this framework, Musk decided to build his own fab, while Intel provides technical and process support to accelerate production. Gelsinger stated he meets weekly with Musk’s team, and progress is smooth.

He also mentioned Musk’s unconventional operational thinking—for example, discussions about allowing smoking in certain cleanroom zones. “I may not go that far, but perhaps in some areas, it could be possible—what matters is maintaining an open mindset,” he said.

Investor Misconception: Intel Is Still in the “Crawling” Phase; True Potential Emerges After 2030

Addressing skepticism about Intel’s transformation pace, Gelsinger invoked his signature “crawl-walk-run” framework. He acknowledged that over the past several months, Intel has remained in the “crawling” phase—building up CPU, GPU, and software architecture teams quietly, aiming to drive leapfrog innovation at the speed of a large startup. On the foundry side, the gap with TSMC remains significant, requiring humility and steady investment in foundational capabilities like IP, yield, and defect control.

“My VC instinct tells me: seek 10x opportunities,” Gelsinger said. Drawing from his experience at Cadence, where he helped generate roughly 76 to 85x returns for shareholders from interim CEO to exit, he admitted Intel’s scale makes replication harder—but his clear goal remains: achieve a 10x return within 5 to 10 years.

Transcript of Interview:

Host: Welcome back to No Priors. Today, Allad and I are joined by Patrick Gelsinger—legendary investor from Walden, former CEO of Cadence, and current CEO of Intel. We’ll discuss his plan to transform Intel, what it means for the U.S. government to become a major shareholder, how to be an exceptional semiconductor investor, and whether we can manufacture chips domestically in the U.S. Welcome, Patrick.

Why Take on This Role at Intel?

Host: Let’s start with a straightforward question. Leading this pivotal American semiconductor company is no small task. Why did you accept?

Gelsinger: A great question. I’m 66 years old—many people say I should retire. Why take on the hottest seat in the industry? Two reasons: first, this is a landmark company vital to the entire semiconductor ecosystem and to America itself. Second, after Cadence, I wanted to do one more big thing.

Host: What surprised you most over the past year?

Gelsinger: The most unexpected event—I had never experienced anything like it in any prior role or training—was when President Trump called me early one morning demanding my resignation due to alleged conflicts of interest, with no exceptions. At first, I convinced myself: I don’t need this job. I’m doing this purely to save Intel. Setting aside personal emotion, I asked: what can I do for Intel? Fortunately, I secured a meeting on Thursday morning and another on Monday. I presented my case: born in Malaysia, raised in Singapore, MIT-educated, lived in the U.S. all my life, never left. He listened—and gave me a chance to continue. I am deeply grateful.

Host: You said this role is about “saving Intel.” What does success look like in your mind?

Gelsinger: I’ve been here 14 months—so much has changed. First, culture change: clear accountability, faster decision-making. I’m used to startup rhythms—everything moves at light speed. But Intel had layer upon layer of bureaucratic meetings—something I had to change. Second, listening to customers: to truly satisfy them, you must be humble, willing to listen, confront their problems head-on. Third, from day one, I decided all engineering teams report directly to me. I’m an engineer—I want to know exactly where things break down and what needs fixing. Listening to customers, satisfying them, ensuring we have the right products, simplifying our portfolio, and building a clear five- to ten-year roadmap and vision.

Intel’s Decade Vision

Host: What’s your vision for Intel in ten years?

Gelsinger: My approach—whether at Cadence or Intel—is always crawl, stay humble, listen to customers; then walk; finally run. Step by step.

First, strengthen the balance sheet—truthfully, it was in bad shape. I’m relieved the U.S. government became a major shareholder. I explained to President Trump: look at Japan, look at Singapore—this is infrastructure-level support, which governments should provide.

Second, I’m deeply thankful to my longtime friend Jensen Huang—he invested $5 billion in Intel, and I’m glad I’ve done meaningful work; that $5 billion has now grown to $25 billion or more. Also, Masayoshi Son from SoftBank—someone I served on the board with—stepped in. Through these partnerships, we’ve stabilized our balance sheet.

Next, focus on products, simplify the portfolio, listen to customers, launch next-generation leading products. Coincidentally, demand for agent AI and inference CPUs is booming—so I’ve caught a favorable moment. Previously, the CPU-to-GPU ratio in training was about 1:8; now I see it shifting to 1:4, even lower. CPUs are regaining importance—I’m pleased.

I’ve spoken with several AI model developers who said CPUs perform better in reinforcement learning phases and coordinating the scheduling of multiple agents. So current CPU demand is very high. After laying a solid foundation in data center server products, another critical business is our foundry. It’s capital-intensive and challenging. You need the right IP portfolio—e.g., low-power IPs for mobile clients. Without these, you can’t serve them. Foundry is a service business—and a trust business. If yield doesn’t meet expectations, customers leave due to revenue loss. That’s why I focus intensely on yield, defect density, cycle time—to ensure high-quality, reliable delivery. Ultimately, we must move toward full-stack solutions—not just silicon. Customers now ask, “Give me the whole rack”—we must offer system-level solutions. I’m steadily advancing these initiatives while recruiting the best talent I can find. By the way, all hiring is done personally—no headhunters involved.

Collaboration with Elon Musk on Terafab

Host: Another widely discussed initiative is Terafab and the collaboration with Elon Musk. Can you explain how this came together and how you’re working together?

Gelsinger: I believe we both agree: Elon Musk is one of the greatest entrepreneurs of this century. We share a common view: semiconductor infrastructure development hasn’t kept pace with AI’s growth—whether in capacity, production efficiency, or power efficiency—there are gaps, and we both see them.

Second, I thoroughly enjoy collaborating with him. He’s unorthodox—he questions every step: “Why use traditional methods?” That’s refreshing. I love hearing different perspectives and jointly finding optimal paths—both sides learn. He has a clear vision: his robots and cars need massive amounts of chips.

Specifically, Terafab involves Musk deciding to build his own fab, and we’re eager to collaborate, helping him accelerate production using our tech and processes—a joint venture. His team is excellent; I meet with them weekly. Working with him is exhilarating. He’s floated ideas like permitting smoking in cleanrooms—while I may not go that far, maybe some zones could allow it. The key is keeping an open mind—we’re seriously considering and evaluating.

Global Semiconductor Supply Chain Shifts

Host: From a macro perspective, how is AI reshaping the global semiconductor supply chain? What observations do you have by country?

Gelsinger: AI’s impact on the landscape surpasses even the internet, and its implications are deeper. AI makes tasks more efficient—thanks to vast numbers of intelligent agents, many tedious manual tasks can now be completed faster. For example, in semiconductor design, timing optimization and time-to-market can be dramatically improved, reducing costs.

AI demand faces several bottlenecks: power constraints—some countries lack sufficient electricity; helium shortages—many underestimate helium’s impact on the industry; and memory shortages—this is the most urgent issue today. Even with expansion, new capacity takes years to ramp up. CPUs and GPUs remain in short supply, prices rise, and costs ultimately pass to end clients.

The companies hit hardest are those resisting AI adoption. AI can boost efficiency across virtually every function. Companies should proactively embrace AI and find better ways to leverage it—whether in forecasting, design, or workload optimization.

Host: The simplest argument against Terafab and Intel’s foundry competitiveness is labor cost and domestic manufacturability. Why continue investing heavily in foundry? What’s the logic?

Gelsinger: When deciding whether to double down on foundry or exit, I heard many voices—too expensive, not feasible. But my conclusion: this is critically important for the U.S., and for the industry as a whole.

We’ve all faced supply chain challenges. Every major semiconductor company must seriously consider supply chain resilience—avoid over-reliance on one or two geographically concentrated suppliers. Increasingly, people realize domestic U.S. manufacturing is essential.

Our most advanced processes, like 18A (equivalent to ~1.4nm), are already planning 1nm and 0.7nm nodes. As process nodes shrink further, line widths become finer than a human hair—extreme complexity. Any error can jeopardize everything. Thus, manufacturing precision becomes an ever-growing bottleneck.

We deeply respect TSMC—we’re excellent partners. The industry needs more capacity to serve customers. So we’re choosing to persevere—long-term, this is pivotal and where I can create greater value for the industry.

Physical Limits and Advanced Packaging

Host: There’s been much discussion about chip scaling hitting physical limits—line widths too narrow to shrink further. When do you think we’ll truly hit a wall?

Gelsinger: We currently have 18A, advancing toward 14A production. I can see paths to 10nm and 7nm—this route is viable, but increasingly costly and difficult. That’s why we need partners, close collaboration with substrate vendors and equipment makers, to jointly improve yield and performance.

Another critical bottleneck is advanced packaging. TSMC has CoWoS; we have EMIB, our next-gen solution. I must ensure it achieves required yields at scale.

When traditional scaling hits limits, I turn to materials for breakthroughs—GaN, SiC, InP—I’ve invested in all three. On packaging materials, I’m focusing on glass—excellent for thermal insulation. I invested in 3DGS. Intel holds around 1,000 patents in modules—how to integrate substrates and modules is a key challenge. Recently, we announced advanced packaging manufacturing collaborations in India and New Mexico. Additionally, I’m exploring synthetic diamond—an exceptional thermal barrier material—I’ve invested in a diamond wafer company.

Engineers’ spirit: you keep hitting bottlenecks, then find ways to cross or circumvent them. Having deep experience across the entire semiconductor lifecycle—from EDA tools to design to manufacturing—I’m now glad to apply these insights to contribute meaningfully to the industry.

Host: Could there be a scenario where process node convergence narrows performance differences among foundries, creating a kind of asymptote?

Gelsinger: Moore’s Law is about doubling transistor density—but power and cost don’t decline proportionally. You can double performance, but area and cost don’t necessarily halve. Unless you discover new materials or new design methodologies. That’s precisely why I’m increasing investment in materials science talent—this is now the core of innovation in the field.

Eighteen years ago, when I was still investing in semiconductors, many top-tier VCs showed zero interest. I remember, after presenting semiconductor opportunities at a partner meeting, half the room made excuses to leave. The rest said, “Do you have any software or services projects?” Eventually, only one or two offered sympathy. Now, NVIDIA under Jensen Huang is valued at $5.3 trillion, Broadcom and TSMC each near $2 trillion, AMD’s Dr. Lisa Su near $800 billion, Intel near $600 billion. Semiconductors are back in vogue—indispensable infrastructure. Fifteen to twenty years ago, almost no VC would join me—only giants like Samsung, ARM, SoftBank. Now, VCs swarm in—investment enthusiasm is sky-high. I’m deeply encouraged.

Challenges in Semiconductor Investing

Host: You’re both a long-term investor and operator. Semiconductor investing presents many hurdles—capital intensity, unpredictable outcomes, deep understanding of workloads, high switching costs for customers, and cyclical nature. How do you assess these risks, and what advice would you give others on where to invest in this supply chain?

Gelsinger: Venture capital and entrepreneurship are in my blood—I genuinely enjoy it. Not to boast, but here’s context: I’ve led 159 IPOs, 126 M&A exits, with over 200 semiconductor investments—38% in the U.S.

My investment approach starts with one core question: Where is the bottleneck? What problem are you solving? For example, I invested in Cradle Semiconductor because interconnects were becoming a bottleneck. I backed Celestial AI because optical interconnects are growing crucial in clusters—Jensen Huang has invested in nearly all photonics-related firms—not a coincidence.

At the design level, can AI and machine learning reduce complexity and enhance design quality? I believe EDA offers enormous opportunity—several startups are moving in this direction, a gold mine. In new materials, GaN, SiC, and InP are my investment focus—some have already been acquired by majors like ADI. Power management—transitioning from 40V to 1V incurs massive losses—another promising bottleneck area.

My investment framework always asks: Is the problem real? Are customers truly struggling with it? Then critically: Who is the first target customer? I prefer hyperscalers—they have capability, willingness, and if they like your product, they’ll spend millions or guarantee volume over the next few years. Winning one big client enables rapid scaling.

Talent is also critical—U.S., Silicon Valley, Austin, Israel are key focus areas. Israel produces highly disruptive, exceptionally driven founders. Even during war, they hold meetings—sometimes saying, “There’s an alert, I need to go to the basement—network might be poor, let’s switch to voice.” Their resilience inspires me deeply.

Today, beyond agent AI, physical AI is the next frontier. One must examine the full stack—why I remain deeply involved in early-stage model investments. I’m extremely bullish on open-source frontier technologies for physical AI—a gold mine.

Cadence Experience

Host: You mentioned AI bringing faster, cheaper, more creative possibilities in chip design and testing. Based on your Cadence experience, which directions are most fertile? Are any already bearing fruit?

Gelsinger: I spent nearly 15 years at Cadence. One of my proudest achievements was finding and personally mentoring my successor—now a superb CEO actively embracing AI, integrating agent AI into tools to boost efficiency. Sassine at Synopsys is doing the same—backed by NVIDIA’s $2 billion investment—and acquiring Ansys to expand into full-system design.

Larger companies are doing it, but there’s room for startups to do more disruptive work—eventually, they can go public or be acquired by two big players. It depends on the founder’s vision. My philosophy: if the founder wants a quick exit, help them achieve it. If they dream of going public from Day 1, guide them through that path. As a VC, we support founders’ dreams and help them realize them.

Scaling and Investment Decision-Making

Host: The areas you mentioned—materials companies, EDA, manufacturing—if viewed 10 years ahead, will Intel or future semiconductor companies be unrecognizable due to AI?

Gelsinger: I believe so. Returning to capital intensity, unpredictability, and cyclicality—these must inform investment decisions. I usually prefer entering early, building the team; finding investors who’ll stick with you through tough times, not just friends who show up in good weather; and seeking strategic investors who add value in manufacturing, storage, interconnects, or other dimensions. I also have contacts in growth and hedge funds—they offer unique perspectives on public markets, helping founders avoid dead ends—very valuable.

Honestly, looking back, of the 10 companies I invested in, 9 changed their business plans mid-stream due to shifting markets. So I prefer founders with strong teams, not solo operators. Open-mindedness matters—willingness to listen, accept feedback, but ultimately form their own judgment. The best outcome isn’t “do whatever you tell me”—it’s when you provide enough input, and they independently derive conclusions you approve or understand. That’s the joy of entrepreneurship.

Looking back in 10 years, winners will be those focused on a niche, found the right partners, and achieved scalability. Full-stack solutions matter—critical. Large companies can follow Jensen Huang’s path—focus on CUDA and platforms, building a dominant platform, which he’s done. Startups can follow Anthropic or OpenAI—reshape the rules elegantly. Startups can move at lightning speed, truly becoming leaders.

For Intel, I hope it plays that role—XPU, advanced packaging, foundry. Integrate them, deliver custom chips for diverse workloads—that’s my direction.

Team Restructuring in the AI Era

Host: Software industry is changing dramatically—what kind of talent to hire, who can manage multiple agents. Many now prefer hiring people aged 30–50, as they’re accustomed to managing teams—this skill transfers directly to managing agents. In hardware or foundry contexts, how do you see team structure and capability evolving?

Gelsinger: Back to the crawl-walk-run framework. In the “crawling” phase, I recruited the best talent in the semiconductor industry. Now I’m assessing what kind of software talent is needed to build full-stack capabilities. I’ve also noticed the average team age is in the 40s to 50s—I need to bring in younger talent to understand workloads and cutting-edge open-source models.

Fascinatingly, my son has become my teacher. Whenever I visit him to play with my grandson, I ask him about AI and machine learning—he knows more than I do. I’ve learned a lot, then translate that into investment decisions and talent acquisition.

Intel was once a very old-school spreadsheet-dependent company. I’m transforming it into an AI-powered enterprise—not just in design, but across the entire organization, reducing reliance on spreadsheets. We’re combining senior technical talent with AI tools—not just in sales and marketing, but actively embracing AI in design.

Industrial Policy and Capital Sources

Host: For capital-intensive enterprises, securing funding is a constant challenge. Industrial policy created giants like TSMC, yet this model has long been unpopular in U.S. business culture. What’s your take?

Gelsinger: For capital-intensive operations and infrastructure projects, access to capital is paramount. Now, some VCs are willing to commit $1 billion to a single company—unthinkable before. So in early-stage strategies, either enter very early at reasonable valuations—or reach Series A, but now Series A valuations exceed $1 billion, making it hard.

Capital that enables scaling—like mutual funds, less sensitive to ownership percentages—I welcome them. For capital-intensive projects like AI factories or foundries, government funding, sovereign wealth funds, or large infrastructure funds are essential. Sovereign wealth and government capital will grow increasingly important.

As a public company, I consciously focus on long-term growth-oriented investors—not short-term traders asking quarterly, “When will you buy back stock?” Shareholder returns are legitimate, but balancing that with building the business is crucial.

Biggest Investor Misunderstanding About Intel

Host: What do you believe investors most misunderstand about Intel today?

Gelsinger: Several points. First, revisit crawl-walk-run: over the past months, I was still crawling—but people are starting to see potential. Product-wise, we still hold market share in PC clients, but performance must improve significantly. So I’m quietly building CPU, GPU, and software architecture teams—preparing for leapfrog leadership, moving fast like a large startup, leveraging better tech to jump ahead.

On foundry, the gap with TSMC remains wide—we must stay humble, focus on fundamentals: IP, yield, defect density, cycle time—to make foundry more efficient and reliable. It’s a trust business—customers must trust you before handing over wafers. These take time, but I believe by 2030 to 2032, people will finally grasp Intel’s true potential.

PC client is our foundation, but we’re expanding into edge, physical AI, and agent AI. Previously, we provided servers and PCs for humans. Now, a new dimension emerges—millions of agents need compute access, software stack access. I believe Intel has opportunity in both agent AI and physical AI—the game isn’t over.

AI is just the beginning—you have Huang’s training side, edge side, agent AI, physical AI—huge opportunity, everyone still has a shot. This is where I’m fully committed. Past 14 months delivered ~6x return to shareholders—but this is just the beginning, with plenty of room to grow.

My VC instinct tells me: chase 10x opportunities. At Cadence, from interim CEO to retirement, the stock rose from $2.4 to ~76x return for shareholders; at exit, likely ~85x. Intel’s scale makes replication harder, but my target is 10x—achieve 10x return within 5 to 10 years. As someone whose DNA is VC, that’s my goal.

Where Will Compute Reside?

Host: Some argue data centers will keep growing larger—gigawatts are just the start—centralization dominates. Yet your business vision also includes edge and client-side computing. How do you see compute distributed between data centers, edge, and client devices—or is it entirely driven by application workloads?

Gelsinger: Current large-scale AI infrastructure build-out is correct. I see no reason for slowdown—workloads keep growing. The current constraints are supply-side: any slowdown comes from supply limitations, not demand.

But I’m more focused on: what applications will run on this infrastructure after it’s built? You must identify scalable applications—just like in the internet era, Amazon and Netflix emerged as true apps, while others faded or were acquired. AI will undergo the same process: massive growth followed by consolidation, eventually revealing a few true winners.

Focusing on applications is key. Netflix and Amazon are real apps—they won. Some applications are better suited for edge or client-side execution—robotics, defense, etc. Device-side compute choices are critical—your assumptions about connectivity and device capabilities determine what you can do. This was often overlooked in the SaaS era.

My investment method: find real problems, identify the right partners, assess whether the application market size is sustainable—if you truly believe it, double or triple down. Of course, this includes betting on applications not yet scaled.

Host: Thank you so much for joining us today—it’s been a genuine pleasure.

Gelsinger: Thank you for having me.

Source: Wall Street Journal