Intel CEO Lip-Bu Tan Signals New RAM Architectures and More Funding Plans

Intel CEO Lip-Bu Tan has reignited discussion about the company returning to memory research and potential future investments, saying the chipmaker is already working on new RAM architectures and is planning additional funding. Speaking on the Techsurge podcast, Tan framed DRAM as something beyond a simple commodity purchase—and hinted that recent staffing moves may be tied to those longer-term memory ambitions.

Tan’s comments: DRAM research as a “pet project”

In the interview, Tan directly addressed speculation around Intel getting more involved in RAM production or development again. While he didn’t offer a firm commitment tied specifically to Intel’s own manufacturing, he made it clear the effort is active:

• Tan said “one of my pet project[s]” involves memory, specifically mentioning new architecture work.
• He referenced hiring Seok-Hee Lee, describing Lee as a “good friend” and noting Lee previously ran SK Hynix.
• He emphasized that memory technology is producing “a lot of new” developments, contrasting it with earlier thinking that avoided memory investment because it was too much of a commodity business.

Crucially, Tan did not specify whether the “man” doing the work is Intel itself or his personal investment activities. However, there is a concrete Intel-related detail in the background: Intel’s separately incorporated foundry business hired Seok-Hee Lee as a vice president in June, and packaging was cited as a key expected focus area.

Intel’s prior memory experience—and what it was built on

Intel is not a newcomer to DRAM-era ideas, but its past involvement has largely been in enabling roles rather than producing memory chips at scale. Historically, Intel founded the commercial DRAM industry and achieved major early successes long before it began modern processor development.

Recent memory-adjacent work and internal patents suggest Intel has continued to explore the building blocks of DRAM systems:

• Intel has worked with SoftBank on “Z-Angel Memory.”
• An Intel patent related to simplifying the connection of stacked DRAM was found under the “XBM” label.
• Intel has pushed MRDIMMs into the market.
• Earlier memory-related initiatives included the HBM competitor Hybrid Memory Cube (with Micron) and, in the early 2000s, RD-RAM (with Rambus).

Even with these efforts, the pattern described here is consistent: Intel’s role was mostly tied to controller and packaging development, while it had not been fabricating memory itself for a long time.

Why this likely won’t fix today’s RAM shortage

Despite the renewed attention, Tan’s own expectations for major foundational investments suggest any meaningful impact won’t arrive quickly. He said that for comparable large-scale investment themes, the company’s relevant time horizon is typically 10 to 15 years.

There are also practical constraints in semiconductor manufacturing:

• Processes designed for logic circuits generally can’t be directly reused for DRAM fabrication, and vice versa.
• Given Intel’s current footprint of factories, the company is unlikely to be able to quickly counter the ongoing RAM shortage in the near term.

In short: Tan’s remarks point to a renewed push in memory architecture research and investment planning, but the timeline for switching from innovation to output is far too long to relieve immediate market pressure.