Tesla-SpaceX Terafab Targets a 2027 Test Chip and Late-2028 Volume Production — Aggressive Schedule and EUV Access Are Key Constraints
On October 20, 2026, Elon Musk said on SpaceX's earnings call that the Tesla-SpaceX Terafab semiconductor venture targets a first functional SRAM test chip in May or June 2027 and volume production in the third or fourth quarter of 2028. The long-term ambition is one terawatt per year of computing power for AI and robotics, but Musk himself warned that the timeline carries 'big error bars' and is extremely aggressive. He did not answer whether EUV lithography tools had been ordered, leaving access to equipment from ASML, Applied Materials and KLA, as well as process yield, as central execution risks. Process node, wafer capacity, capex, ownership and Tesla's funding share remain undisclosed, so the 1 TW goal cannot be translated into output or revenue. Success could vertically integrate Tesla's AI and robotics chip supply, improving long-run cost and control, but it is too early to assume a material reduction in reliance on NVIDIA or AMD before 2028.
A. Key facts
On SpaceX's Q3 2026 earnings call on October 20, Elon Musk said the Tesla-SpaceX Terafab venture aims to produce its first functional SRAM test chip at its Austin research facility around May or June 2027 and enter volume production in the third or fourth quarter of 2028.
Musk explicitly described the schedule as extremely aggressive and subject to major uncertainty. It is therefore a management target, not a firm commercial-production commitment.
The Texas project is intended to integrate advanced logic and memory manufacturing, advanced packaging and testing in one complex and ultimately target one terawatt per year of computing power for AI and robotics.
Intel announced cooperation with the project in April 2026 and potential cooperation with TSMC has been discussed, but Musk has recently said Tesla and SpaceX intend to operate the facility themselves.
B. Business significance · InveDeck interpretation
The event matters to Tesla because it provides the first concrete timeline for potentially verticalizing AI chips used in autonomy, Optimus and data centers from design through manufacturing and packaging. If successful, it could improve supply assurance, process customization and long-run unit-cost control.
The counterweight is that an advanced fab requires years of heavy upfront capex and typically suffers low initial yields. Direct operation by Tesla and SpaceX would internalize foundry, equipment, talent and process-integration risks.
This is a separate operating roadmap from the possible Tesla-SpaceX merger review, but larger shared assets and funding obligations increase the importance of valuation, conflicts controls and capital-allocation scrutiny in any future transaction.
C. Quantifiable impact
The disclosed quantitative milestones are a first SRAM test chip in May-June 2027, potential volume production in Q3-Q4 2028 and a long-run target of one terawatt per year of computing power.
The 1 TW figure is a computing-power ambition reported by Reuters, not a disclosed electricity load, wafer-start capacity, chip count or revenue figure. Without process node, wafer starts, yield or ASP, it should not be converted into production or sales.
Total project capex, Tesla and SpaceX funding shares, equipment deposits, depreciation and breakeven timing are undisclosed, preventing a defensible estimate of FCF, leverage or margin effects.
D. Link to earnings and expectations
Tesla has secured $30 billion of undrawn credit facilities to support a 2026 capex plan above $25 billion centered partly on AI and data centers, but it has not disclosed any amount allocated to Terafab.
The new comments move the advanced-fab concept disclosed in March 2026 from a broad ambition to an execution roadmap with test-chip and volume-production milestones. The explicit uncertainty warning means this is neither a guidance increase nor confirmed supply.
Tesla's next earnings disclosures should clarify whether Terafab spending is inside its existing 2026 plan and how sharply the 2027-2028 investment curve may rise.
E. Industry and supply chain · InveDeck interpretation
If successful, some Tesla-SpaceX AI and robotics chip demand could eventually shift in-house, strengthening purchasing leverage with NVIDIA, AMD and external foundries. A 2028 production target and initial-yield risk do not support a near-term GPU-demand reduction thesis.
The project could create incremental demand for ASML EUV tools, Applied Materials and KLA process and inspection equipment, advanced packaging and memory. Musk's refusal to directly confirm EUV orders highlights tool-slot access as a critical bottleneck.
Combining advanced logic, memory and packaging in one facility is a technically difficult departure from the industry's specialized supply chain. Process yield, qualification and the design ecosystem may matter more than equipment installation alone.
F. What to watch next
- Actual functionality, process node and yield of the planned May-June 2027 SRAM test chip
- Orders, delivery schedules and deposits for ASML EUV and Applied Materials-KLA equipment
- Ownership, IP and output allocation among Tesla, SpaceX and Intel, including related-party controls
- Total 2027-2028 capex, Tesla's funding share, depreciation and financing plan
- AI5/AI6, Optimus and SpaceX AI chip tape-outs and the share still sourced from external foundries
Sources and verification scope
- Reuters · 2026-10-20
Reuters directly reported the test-chip and volume-production targets, stated uncertainty and equipment-procurement constraints disclosed by Elon Musk on SpaceX's Q3 2026 earnings call.
- Reuters · 2026-03-22
Earlier reporting describing the joint-development background and the companies' internal AI and robotics chip demand.
- Reuters · 2026-10-07
Recent reporting on Intel cooperation, discussions with TSMC and Tesla-SpaceX's intention to operate the facility themselves.
Reported facts are separated from InveDeck interpretation. Read the original reporting through the source links.
Revision history (1)
- 21/10/2026, 09:26 KST · Revision 1 — Initial publication: incorporates the first specific 2027 test-chip and late-2028 volume-production targets disclosed on SpaceX's earnings call.