Bernstein "Pours Cold Water": Musk's Mega Chip Factory Requires $5-13 Trillion in Capital Expenditure, "Harder Than Landing on Mars"

Elon Musk’s ambitious “Terafab” plan faces a cool assessment from Wall Street.

According to ZF Trading, Bernstein Research’s latest report provides detailed quantitative estimates indicating that to achieve Musk’s proposed annual 1 terawatt of computing power production, capital expenditures would need to reach $5 to $13 trillion, with wafer capacity equivalent to the current global semiconductor capacity—making this challenge “harder than landing on Mars.”

Last weekend, Musk announced the launch of the “Terafab” project, aiming to expand human computing power production to 1 terawatt annually, about 50 times the current global supply (around 20 gigawatts). The plan includes building an advanced wafer factory in Austin, Texas, covering the entire supply chain for computing chips, logic chips, memory, packaging, and mask templates. Led by Bernstein analyst Stacy A. Rasgon, the team released a report systematically evaluating the feasibility of this plan.

Bernstein pointed out that if investors truly believe Musk can achieve this goal, semiconductor equipment stocks would be the most direct beneficiaries; however, as of now, the plan’s actual impact on the industry “may not be significant, mostly hype.”

Astronomical Capital Expenditure Estimates

Using NVIDIA’s rack-based architecture as a baseline, Bernstein roughly estimated wafer demand for GPU, HBM memory, and CPU chips. The report shows that producing 1 terawatt of power annually would require monthly startup volumes of 7 to 18 million 300mm wafers, with HBM memory demand dominating.

Converting to a factory capacity of 50,000 wafers per month, this demand is equivalent to building 140 to 360 such factories. At a capital expenditure of $35 billion per factory, the total investment would reach $5 to $13 trillion. Bernstein explicitly notes that these estimates are “very rough” and do not include other semiconductor categories like network, optical, analog, and power chips.

Capacity Needs Comparable to Rebuilding the Global Semiconductor Industry

Bernstein’s estimates reveal the staggering scale of this plan’s capacity requirements. The report states that the wafer capacity needed for 1 terawatt of power is roughly equivalent to the entire current global semiconductor capacity (about 16 million 300mm wafers per month).

Focusing only on “related” semiconductors—i.e., memory plus advanced logic wafers below 4nm—the current global installed capacity is about 5 million wafers per month. The 1-terawatt goal would require several times that capacity. In other words, Musk’s plan essentially demands multiple expansions of existing advanced process capacity, which is extremely challenging.

Industry Impact: Equipment Stocks Benefit, Foundry Model Unthreatened

Bernstein believes that the Terafab plan will have limited immediate impact on the semiconductor industry but offers several noteworthy investment directions.

The report states that if the market believes Musk can push this plan forward, semiconductor equipment (semicap) stocks will be the most directly benefited.

Regarding whether Musk’s in-house chip manufacturing could threaten existing chipmakers, Bernstein remains relatively optimistic, suggesting that in an environment with such strong demand for computing power, any participant will face significant upward potential beyond their current capacity, including memory manufacturers.

On a business model level, Bernstein notes that Terafab would integrate logic, memory, mask manufacturing, chip design, and packaging into one entity, essentially forming a “super IDM” model. However, this model has been proven less efficient than the division of labor seen in “foundry + fabless + specialized memory IDM” systems.

Therefore, Bernstein concludes that Terafab currently poses limited threat to foundries like TSMC.

Bernstein: Not dismissing Musk, but the challenge is “extremely daunting”

Despite the astonishing estimates, Bernstein does not completely rule out Musk’s success.

The report states that Musk has previously achieved many things that others considered impossible, “we won’t dismiss him lightly.”

However, it also clearly notes that “a true Terafab, in our view, feels more like an extension, especially under the current computing paradigm.”

Bernstein suggests two alternative paths:

First, if Musk cannot advance independently, he might seek partnerships with existing chip manufacturers;

Second, Musk might have some “more surprising” technological approach to break through the current paradigm, but the report admits, “we don’t know what that might be.”

This insightful content is from ZF Trading.

For more detailed analysis, real-time insights, and frontline research, please join 【**ZF Trading Annual Membership**】

![](https://img-cdn.gateio.im/social/moments-fc48d52a1d-cfade68ac3-8b7abd-ceda62)

Risk Disclaimer and Terms of Use

Market risks exist; investments should be cautious. This article does not constitute personal investment advice and does not consider individual users' specific investment goals, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. Invest at your own risk.