In June 2026, Tesla hired Gary Jiang, who spent about 18 years at Intel leading its 18A process ramp, as "Director, Tera Fab." He will work full-time on the Terafab project in Austin, Texas, from June 2026.
March 21, 2026 · Tesla · SpaceX · xAI
Tesla Recruits an Intel 18A Veteran to Build the Trillion-Watt "Terafab"
Gary Jiang — ~18-20 years at Intel in fab operations, tied to the 18A and 14A nodes — joins as Director, Tera Fab to turn Musk's joint mega-fab vision into real production capacity.
1 TW
Target AI compute output per year
~50×
vs current global AI compute (~20 GW/yr)
65%
Intel 18A yield, stabilized in Arizona
$5T+
Bernstein's full-build cost estimate
The leap in scale: annual AI compute
Same unit (GW/yr). Each block ≈ 100 GW.
~20 GW
Current world
1,000 GW (1 TW)
Terafab
≈ 50× today's entire global AI compute production.
Inside the build
Process
Intel 18A (~1.8nm, GAA + Backside Power), 14A planned next
Packaging
Chiplets via EMIB + Foveros 3D stacking
Wafer scale
Target 1 million wafers / month at full scale
Sites & timing
Near Austin — proto 2026, mass production ~2027
Why it could work
An 18A/14A manufacturing veteran boosts execution
Leverages Intel's existing fabs + packaging — not from scratch
Vertical integration may speed development cycles
Strengthens U.S. onshoring, backed by the CHIPS Act
Why skeptics doubt it
ASML EUV tools prioritized to TSMC, Samsung, Intel
Memory manufacturing needs separate expertise
Full build could exceed $5 trillion
Heavy reliance on Intel; realism as TSMC rival unproven
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…