Recently, Intel’s current CEO Lip-Bu Tan accepted an exclusive interview with the technology podcast No Priors, revealing many investment directions and demonstrating Intel’s ambitions for future development.
Chen Liwu has set a goal of creating a 10-fold return for shareholders within 5 to 10 years.
He predicts that Intel’s business potential will be fully unleashed from 2030 to 2032, when the market will truly realize its value.
As of the close of trading on June 18, Intel rose 10.64% to a new all-time high, with a total market value of US$673.434 billion. Since the beginning of the year, Intel’s stock price has risen by more than 263%.
66-year-old Chen Liwu took over Intel in a crisis.
He originally planned to retire and spend time with his family, but when Intel offered him his first invitation in 2021, he declined because of his commitment to Cadence’s successor.
When he was invited again in 2025, 80% of his old friends in the industry advised him not to accept it, believing that failure would ruin his lifetime reputation and be extremely risky.
But the words of an old customer friend moved him, saying that Intel is the cornerstone of the semiconductor industry and a core enterprise of the US strategy. He has the ability to save it and should complete this major task before retiring.
After taking over Intel, the first thing Chen Liwu did was to completely repair the poor balance sheet.
Without healthy cash flow, all technological investment and production expansion are empty talk.
The U.S. government has become an important major shareholder of Intel. Semiconductor manufacturing is a national strategic infrastructure, and national support is essential.
Huang Renxun invested US$5 billion in real money in Intel, driving upstream and downstream supporting funds, and overall leveraging more than 25 billion resources.
As an old ally, SoftBank’s Masayoshi Son also provided key financial support.
After stabilizing the finances, Chen Liwu immediately made drastic changes to the interior.
He cut off multiple layers of bureaucracy and established a direct accountability mechanism. All core engineering teams reported directly to him; he required all employees to synchronize negative customer feedback 24 hours a day. If the customer told him the problem first, the corresponding person in charge would be held accountable; implemented rigid tape-out rules, and the first and second tape-out of the chip must reach
For standard mass production, the person in charge will be directly replaced if tape-out fails three times or more; the engineer’s status will be raised to the highest level, and he will cross the eighth and ninth floors to investigate the actual situation of the technology; the entire company will undergo a comprehensive AI transformation, eliminate the old spreadsheet-driven management model, and embed AI tools in everything from R&D, production, and finance.
In the past 14 months, Intel shareholders have received a return of about 6 times the market value, but Chen Liwu made it clear that this is just a warm-up.
Chen Liwu proposed in the interview that CPUs are being revived and are not being replaced by GPUs.
With the rise of Agentic AI and inference workloads, the structure of computing needs is changing.
In the training phase, the CPU:GPU ratio evolves from 1:8 to 1:4; in the inference phase, the ratio further moves closer to 1:1; in reinforcement learning and agent orchestration, the CPU is actually better than the GPU.
This means that the value of the CPU has been greatly underestimated in the entire AI process. Intel, as the leader of the x86 architecture, is repositioning its role in AI infrastructure.
In addition, Intel’s ambition is not only to sell chips, but also to provide full-stack solutions, including software and even entire rack systems.
Chen Liwu also revealed that Intel is advancing the Terafab project with Musk, which stems from the joint judgment of both parties that the development of semiconductor infrastructure lags behind the growth of AI demand in terms of production capacity, production efficiency, and power consumption efficiency.
Under the cooperation framework, Musk decided to build his own wafer factory, and Intel will provide technical and process support to help accelerate production, and the cooperation is progressing smoothly.
In terms of technology layout, Intel has fully laid out the 18A and 14A mass production routes, and the team is even planning the follow-up routes for 1 nanometer and 0.7 nanometer.
However, the further down the process node goes, the higher the research and development costs and the greater the technical difficulty. Sooner or later, traditional size reduction will reach physical bottlenecks.
Therefore, while Intel is pushing forward the process, it regards advanced packaging as the next breakthrough. The mass production yield of the EMIB solution is a hard nut that must be overcome at the moment.
Chen Liwu is turning his focus to advanced packaging technology EMIB, glass substrates, and new materials such as gallium nitride (GaN), silicon carbide (SiC), indium phosphide (InP), and synthetic diamond to cope with the challenges of traditional process node shrinkage approaching physical limits.
Regarding the supply chain, Chen Liwu emphasized the strategic value of advanced manufacturing in the United States and believed that no large semiconductor company could concentrate its supply chain in one or two geographical areas.
However, it is still uncertain whether the rising manufacturing costs and lengthened collaboration cycle caused by decentralized layout will eventually be passed on to downstream companies and ordinary consumers.
For foundry, Chen Liwu locked his priority indicators into yield, defect density and cycle time. It is expected that by 2030 to 2032, the true potential of Intel’s foundry business will be reflected in the market, and will not be limited to the traditional basic market of PC clients, but will also extend to emerging markets such as edge computing, physical AI and intelligent AI.
