Elon Musk's ambition to manufacture chips is accelerating. Intel's CEO, Lip-Bu Tan, sent an internal memo to employees last week promising to disclose the "scope and nature" of the company's involvement in the Terafab project within "weeks," signaling a major step towards the public realization of this grand plan aimed at reshaping the semiconductor manufacturing landscape.
According to the memo obtained by CRN, Lip-Bu Tan communicated this information to all Intel employees last Friday. This timing comes just two days after Intel formally announced its partnership with Musk's SpaceX, xAI, and Tesla to advance the Terafab project. Intel will be deeply involved in the project as a foundry services provider, a move seen as a significant element of its own transformation and revitalization strategy.
Terafab's near-term goal is to establish a pilot production line in Austin, Texas, with plans to commence silicon wafer manufacturing in 2029, starting with an initial capacity of 3,000 wafers per month.
On the supply chain front, Bloomberg previously reported that the Terafab team has sent requests for quotes to equipment giants like Applied Materials, Tokyo Electron, and Lam Research, demanding responses at "light speed." The project's ultimate vision is to achieve an annual supply of 1 terawatt of computing power, directly challenging TSMC's dominance in advanced process technologies. Analysts at Bernstein estimate the total capital expenditure required for the project's full implementation could range from $5 trillion to a staggering $13 trillion.
**Intel Memo: Details of Involvement to be Revealed Soon**
According to the CRN report, Lip-Bu Tan's memo explicitly stated that Intel would disclose the "scope and nature" of its participation in Terafab to employees but did not provide further specifics on the exact timing.
Previously, Musk visited Intel's headquarters in Santa Clara. Following the visit, Lip-Bu Tan posted a photo of their meeting on social media; Intel board member Ira Ehrenpreis was also present. This is the most direct public record of contact between Intel's senior leadership and the Terafab team confirmed to date.
Intel has characterized this collaboration as a vital component of its efforts to revitalize its foundry business. The company has recently been focused on establishing Intel Foundry as an independent manufacturing platform capable of handling orders from external clients and actively seeking endorsements from major customers, including Nvidia. Terafab's involvement is seen as a key moment that adds substantive weight to this strategy.
**Terafab's Two-Phase Blueprint: Targeting Automotive, Robotics, and AI Chips**
Information disclosed by CRN indicates the Terafab project will advance in two phases. The first phase will focus on constructing a wafer fabrication facility for automotive and robotics chips, while the second phase will shift to mass production of AI chips.
In terms of end applications, the chips produced are intended primarily for Musk's xAI artificial intelligence business, the Optimus humanoid robot, and the Robotaxi project, while also catering to SpaceX and xAI's space data centers. Musk has indicated that xAI is expected to consume the vast majority of the output.
Regarding production capacity, the project plans to start a pilot line in Texas in 2029, with an initial output of 3,000 wafers per month, which is expected to increase as the project scales. The project also aims to internalize the entire process flow, from mask production to testing and packaging, and has begun talent recruitment, reaching out to engineers from companies like Applied Materials, Samsung, and TSMC.
Tesla's recently disclosed A15 chip is currently manufactured by TSMC and Samsung, but moving its production into Terafab's own facilities is considered a primary goal once the project is operational. Furthermore, Tesla has confirmed it is developing a next-generation A16 chip and the Dojo3 supercomputing project.
**Capital Hurdles and Technical Path Remain Key Uncertainties**
Despite the noticeably accelerated pace, skepticism about Terafab's feasibility persists. Bernstein analysts estimate the capital expenditure required for full implementation could be between $5 trillion and $13 trillion, potentially making it the largest single investment concept in semiconductor industry history.
As previously reported by Bloomberg, Tammy Qiu, Head of Technology Equity Research at Berenberg Bank, stated that "the intent of the project is real" but does not expect substantial progress within the next two years. She noted that Berenberg has not yet incorporated Terafab into its financial models for ASML. As the sole global supplier of extreme ultraviolet (EUV) lithography machines, ASML is indispensable for any manufacturer aiming to mass-produce advanced chips. However, it remains unclear whether the Terafab team has initiated contact with the Dutch company.
Currently, Terafab has not placed any firm orders, and its chosen technology path and final production locations are yet to be determined. It is also still unclear whether the project will expand to multiple sites beyond Texas. Bloomberg reported that after receiving a collaboration request from Terafab, Samsung did not respond directly but instead proposed an alternative: allocating more foundry capacity for Tesla within its planned factory in Taylor, Texas.
With Intel's CEO promising to publicly disclose more details within weeks, the market is set to receive the most comprehensive roadmap of Intel's involvement in Terafab to date. Whether this disclosure will alleviate external doubts about the project's viability remains to be seen.
Comments