Tesla's growth narrative is undergoing a fundamental shift: while batteries and vehicles remain sources of cash flow, the next phase of imagination is pinned on physical AI—Robotaxi, Optimus, and AI models and computing stacks for shared human-vehicle interaction. The bottleneck is also shifting: it's not battery production capacity, but rather chip supply, manufacturing capability, and supply chain security under geopolitical pressures.According to the Chase Trading Desk, Barclays' U.S. automotive and mobility analyst Dan Levy bluntly stated in a recent research report: &#34;Chips will become the pillar of Tesla's next phase of growth,&#34; and to push this growth to &#34;mega-scale,&#34; Tesla needs a Terafab.Barclays views Terafab as Tesla's next round of vertical integration's &#34;extreme move&#34;: to create a closed loop for logic chips, storage, and advanced packaging as much as possible in the U.S., aiming to cover the AI6/AI7+ required for Robotaxi/Optimus, as well as Dojo-related chips on the data center side.The problem is that once the scale is opened according to Tesla's standards, the bills will also become frightening: Barclays estimates that even at a lower target, capital expenditures will need to be around $20 billion to $25 billion, while more aggressive capacity plans could raise it to $40 billion to $50 billion. Bank of America bluntly stated that the Terafab project faces a capital expenditure threshold of over $60 billion, and even in the most ideal scenario, its 2nm wafer pricing will still be higher than TSMC, making investment returns difficult to rationalize.More critically, there is the execution path. Tesla lacks experience in large-scale chip manufacturing; leading processes, tool chains, yield rates, and packaging are not problems that can be solved simply by &#34;building a factory.&#34; Compared to going solo, Barclays prefers a &#34;cooperative version of Terafab&#34;—Tesla funds to secure capacity, while the foundry is responsible for manufacturing, with potential partners including Samsung, TSMC, and even Intel.<h2>What exactly does Terafab aim to do? A domestic closed loop for logic, storage, and packaging</h2>The research report pulls the timeline back to Battery Day in 2020: at that time, Tesla used battery production capacity as a &#34;strategic pivot for the next decade.&#34; Now, in the new phase driven by Robotaxi and humanoid robots, Barclays believes the new pivot has become chips—they directly determine whether vehicle-side reasoning, robot-side reasoning, and the data center training/validation chain can expand.Musk emphasized multiple times at the shareholder meeting in November 2025 and the Q4 2026 conference call that in the next 3-4 years, chips (including AI logic and storage) will become limiting factors for growth. Recently, he also mentioned on X that Terafab will launch in a few days, prompting the report to discuss Terafab as a time window for &#34;the next big move.&#34;The ambition of Terafab goes beyond expanding production; it aims to create a complete closed loop for &#34;logic + storage + advanced packaging&#34; in the U.S.<blockquote>In terms of logic chips, Tesla has accumulated considerable design experience from HW3 to AI6 and has established manufacturing anchors at TSMC in Arizona and Samsung in Texas, making this the relatively mature part among the threeIn terms of packaging, Tesla has accumulated experience in design, but has limited manufacturing experience—packaging precisely determines whether logic and storage can be tightly coupled to achieve higher energy efficiency, which is a key variable in the performance of the entire computing unit.Storage is the biggest blank: Tesla has neither design experience nor manufacturing experience, and the supply of advanced storage in the U.S. is also severely lacking—Micron's Idaho factory will not have initial output until mid-2027, and several factories in New York are not expected to start production until 2030.</blockquote>This makes the difficulty of &#34;full-link localization&#34; not linearly increasing, but multiplicatively compounded.Behind Terafab, there are two strategic motivations that cannot be ignored. The first is geopolitical de-risking: potential disturbances in the Taiwan Strait and U.S.-China trade tensions keep the logic chip supply chain reliant on TSMC hanging by a thread, while the lack of advanced storage supply in the U.S. further increases the attractiveness of &#34;full-link localization.&#34; The second is design control: Tesla hopes to couple logic and storage more closely through packaging, reducing discrete components, and creating computing units truly customized for its own software stack—research reports cite management's expectations that the performance of AI5 could rival NVIDIA's Blackwell (Thor), but with only one-third of the power consumption and costs below 10% of theirs.This unified strategy of &#34;one chip, multiple scenarios&#34; will converge vehicle-side reasoning, Optimus reasoning, and some data center applications all onto self-developed chips, making AI5/AI6 the hub connecting the three business lines of vehicles, robots, and data centers.Research reports estimate that Tesla will procure about 3 million to 4 million chips in 2025. In other words, if the demand curve climbs at this slope, suppliers need to prepare not for &#34;small expansions&#34; but for a round of high-risk capital investment. Musk has mentioned a vision of 160,000 WSPM (wafer starts per month), and Barclays believes this could correspond to about 24 million chips per year under good yield conditions.<h2>$20 billion to $50 billion is just the starting point; economically, it may not be sustainable!</h2>The cost of the vision is a frightening bill.The research report provides two cost ranges: covering about 12 million chips per year requires about $20 billion to $25 billion in capital expenditure; if the target rises to 24 million chips per year, it will require $40 billion to $50 billion. The reference point is: Samsung's Taylor project announced $17 billion corresponding to about 20,000 WSPM capacity, TSMC's total investment in Arizona disclosed to be $165 billion, and Micron's planning of $100 billion to build multiple storage factories in New York. Musk's mentioned vision of 160,000 WSPM is close to the scale of &#34;multiple large factories combined&#34;—a typical wafer fab has about 20,000 to 40,000 WSPM.The sharper question is: These investments are not reflected at all in Tesla's existing 2026 capital expenditure guidance ($20 billion). Barclays itself has already predicted Tesla's free cash flow for 2026 to be -$3 billion, and if wafer fab-level investments are added, the cash flow pressure will be magnified several timesFor this reason, the research report holds a highly reserved attitude towards &#34;independent factory construction.&#34; Tesla lacks large-scale chip manufacturing experience, and the process accumulation, EUV equipment cycle, and advanced packaging complexity required for leading processes cannot simply be resolved by &#34;building a factory.&#34; The report even cites the cancellation of the Dojo project and the impairment provision, as well as the underwhelming production and performance of the 4680 battery, as cautionary tales, reminding the market not to equate &#34;design capability&#34; directly with &#34;manufacturing ramp-up capability.&#34;On March 23, Bank of America Merrill Lynch released a research report stating that the Terafab project faces a capital expenditure threshold of over $60 billion. Even in the most ideal scenario, its 2nm wafer pricing will still be higher than TSMC, making the investment return difficult to justify.Specifically, Bank of America indicated that under the best assumptions of 100% utilization and yield, the fixed cost of Terafab's front-end wafers is approximately $6,000, which is still 1.3 to 1.5 times higher than TSMC's advanced nodes. To maintain the approximately 45% gross margin required for the technology and capacity roadmap, Terafab must price its 2nm wafers at around $32,000, higher than TSMC's level of about $30,000.Even without considering technical challenges, the construction of facilities alone requires 3 to 5 years: it takes about 1.5 to 2 years for the factory to be ready, about 1 year for equipment installation, and 1 to 2 years for risk production and product certification. The report estimates that if construction were to start today, Terafab would not achieve mass production until at least 2029.<h2>Few precedents for going solo: A more likely version is &#34;Tesla pays, giants build&#34;</h2>Barclays prefers a &#34;cooperative version of Terafab&#34;: Tesla funds to secure capacity, while the wafer fab is responsible for manufacturing, encouraging suppliers to commit to more aggressive capacity ramp-up through capital investment and loss guarantees.Among potential partners, Samsung is seen as the most natural choice—covering logic, packaging, and DRAM simultaneously. The report even suggests that Tesla could intervene to some extent in the planning of its second factory in Taylor; TSMC is also a natural partner but does not produce automotive-grade DRAM; Intel aligns with Musk's preference for the American supply chain and may have available capacity.For the stock price, Barclays maintains a cautious stance: the grand goals for chip capacity can continue to fuel long-term growth narratives, but once it comes down to capital expenditure and free cash flow, the market will need more concrete execution details.Barclays maintains an Equal Weight rating on Tesla, with a target price of $360, implying about a 5% downside compared to the closing price of $380 cited in the report—while the chip story can be bigger, until the executable path and cost boundaries are clear, Terafab remains a &#34;high-risk, high-reward, more likely to be collaborative&#34; open proposition

TSLA

TSLQ

Barclays believes that Tesla's growth narrative is switching tracks—batteries are giving way to chips, and the importance of supply chain security is increasingly significant under geopolitical tensions. The ambitions for Robotaxi and Optimus are giving rise to "Terafab": creating a closed-loop system for logic, storage, and packaging in the U.S. Barclays estimates capital expenditures could reach between $20 billion and $50 billion, while Bank of America points directly to a $60 billion threshold, arguing that the economics are difficult to justify. There are almost no precedents for building factories independently; "Tesla pays, and the giants come to build" may be a more realistic outcome

- Tesla's growth narrative is shifting towards AI technologies, with a focus on Robotaxi, Optimus, and shared AI models, while cash flow continues from battery and vehicle sales.  
- Analysts from Barclays emphasize that chips will be crucial for Tesla's next growth phase, requiring significant capital investment, estimated between $20 billion to $50 billion, for the development of the Terafab project aimed at localizing chip production.  
- The challenge lies in Tesla's lack of large-scale chip manufacturing experience, leading to suggestions for collaborative efforts with established chip manufacturers like Samsung and TSMC to ensure capacity and capability for future needs.

Detailed Explanation of Terafab – Why Musk Must Build a "Chip Factory"