On August 6 the Tesla Megapack account announced that Megapack 3 is rolling off the line at the Brookshire, Texas Megafactory — sixteen months from groundbreaking — with a designed capacity of 50 GWh a year dedicated to Megapack 3.
That 50 GWh figure comes from the company's product account, not from a filing: Tesla's Q2 2026 update lists Texas Megapack in its installed-capacity table with the status "Commissioning" and no capacity number against it. Everything below treats the 50 GWh as the company's stated design target and everything else as filed fact.
Here is what it is worth, and what has to be true for it to be worth that.
What the energy business actually is right now
From the Q2 2026 update, five quarters of the segment:
| $M | Q2 2025 | Q3 2025 | Q4 2025 | Q1 2026 | Q2 2026 |
|---|---|---|---|---|---|
| Energy revenue | 2,789 | 3,415 | 3,837 | 2,408 | 3,139 |
| Energy cost of revenue | 1,943 | 2,342 | 2,739 | 1,456 | 2,499 |
| Energy gross profit | 846 | 1,073 | 1,098 | 952 | 640 |
| Energy gross margin | 30.33% | 31.42% | 28.62% | 39.53% | 20.39% |
| Storage deployed (GWh) | 9.6 | 12.5 | 14.2 | 8.8 | 13.5 |
Energy revenue was $3,139M, up 13% year over year, on 13.5 GWh deployed, up 41%. Trailing-twelve-month deployments are 49.0 GWh — a record, as the update says, and the number that matters for everything that follows.
The Q2 gross margin of 20.39% is not the run rate in either direction. Management attributed it to a roughly $240M warranty true-up on vendor cell issues in legacy deployments, plus the non-repeat of more than $200M of tariff benefits that had flattered Q1's 39.53%. Strip both and the segment sits around the 28–31% where it spent all of 2025. We use 28% below, deliberately at the low end, because the update also says quiet-part-out-loud: "average selling prices for industrial storage are coming down amidst growing competition."
The unit that makes this estimable
Energy revenue divided by GWh deployed: $3,139M ÷ 13.5 GWh = $232.5M per GWh, or about $233 per kWh.
That is a blended figure — the segment includes Powerwall and solar, not only Megapack — so it overstates pure Megapack pricing somewhat. Against falling industrial ASPs and a Megapack 3 explicitly engineered for lower installed cost (Megablock claims 40% lower construction cost and 23% faster installation), we run the scenarios at $200 per kWh and show the sensitivity.
Capacity is not the constraint. Demand is.
Tesla's own installed-capacity table puts Megapack at 40 GWh in California and 20 GWh in Shanghai — 60 GWh a year of existing nameplate. Trailing deployments are 49.0 GWh. So the existing plants are running at roughly 82% of nameplate.
Adding 50 GWh takes Megapack nameplate to 110 GWh a year — more than double what Tesla deployed in the last twelve months. The update's own outlook section is explicit that this is the binding question: "Installed capacity ≠ current production rate," and "Deliveries and deployments will be impacted by aggregate demand for our products, supply chain readiness and allocation decisions."
This is the whole estimate in one sentence: the factory does not create $10B of revenue, it removes the ceiling on $10B of revenue. Whether the revenue arrives depends on grid-storage demand roughly doubling, which is a market question, not a manufacturing one.
Three scenarios
All at $200/kWh and a 28% segment gross margin, against 3,528M diluted shares, assuming ~70% of incremental gross profit reaches operating income (the plant is already in the capex base; opex scales sublinearly) and an 18% tax rate. These are our estimates, not Tesla guidance.
| Texas GWh 2027 | Texas GWh 2028 | 2028 revenue | 2028 gross profit | 2028 EPS effect | |
|---|---|---|---|---|---|
| 1. Demand-limited | 12 | 25 | $5.0B | $1.40B | ~$0.23 |
| 2. Base | 20 | 38 | $7.6B | $2.13B | ~$0.35 |
| 3. Nameplate | 30 | 50 | $10.0B | $2.80B | ~$0.46 |
For scale: Tesla's trailing non-GAAP EPS across the five quarters above is about $1.74. So the Texas line is worth 13% to 26% of current earnings power once mature — meaningful, and not the thing that re-rates the stock on its own.
At the current blended $233/kWh rather than $200, scenario 3 becomes $11.6B of revenue and roughly $0.53 of EPS. At $170/kWh it becomes $8.5B and $0.39. The ASP assumption moves the answer by about 30%, which is more than the ramp assumption does — and ASP is the variable Tesla controls least.
Why the ramp assumption is the conservative part
Lathrop took roughly two years to approach its 40 GWh nameplate. Shanghai is at 20 GWh installed and, per the update, still ramping. A sixteen-month build is genuinely fast; a sixteen-month ramp would not be. Production starting in August 2026 makes 2027 the first full year and 2028 the first plausible year at or near nameplate — which is how the table above is laid out.
The counterweight is that Megapack 3 and Megablock were designed for this factory rather than retrofitted into it, and Megablock ships four Megapack 3 units pre-integrated with transformer and switchgear. Fewer site-level integration steps is exactly what shortens the gap between what a plant can build and what a customer will accept.
What it does to the Rule of 40, in both directions
Tesla's Rule of 40 score for Q2 2026 is 21.6 — 25.52 of revenue growth minus 3.87 of free-cash-flow margin. Last quarter it was 22.23. The cash half is negative because capital expenditure was $5,789M in the quarter, up 142% year over year, pushing free cash flow to −$1,092M.
Megafactory Texas is inside that number. Which means the honest way to read this announcement against our metric is:
- The score is being paid for now. The capex that built the plant has already gone through the cash line and dragged the score down for several quarters.
- The score gets it back later. Once the plant deploys rather than builds, the same asset shows up as revenue growth and as free cash flow, on a segment with a ~28% gross margin against Tesla's 16.8% company-wide.
A company whose Rule of 40 sits at 21.6 while it spends $5.8B a quarter is not a company with a broken score. It is a company mid-build. The test is whether the deployments arrive before the depreciation does.
What would falsify this
Three things, in order of how much they would move the estimate:
- ASPs falling faster than the cost curve. The update already flags competitive pressure on industrial storage pricing. If the realised figure lands nearer $150/kWh, scenario 3 is a $7.5B line, not $10B.
- Deployments not doubling. 110 GWh of nameplate against 49 GWh of trailing deployments needs the demand side to more than double. If it does not, Texas cannibalises Lathrop's utilisation instead of adding to it, and the incremental revenue is a fraction of the table above.
- Energy gross margin normalising below 28%. Two of the last five quarters were distorted by one-offs in opposite directions. If the clean number is 24% rather than 28%, every gross-profit and EPS figure above comes down by about a seventh.
The number to watch is not the factory. It is quarterly storage deployments against the 60 GWh of nameplate Tesla already had — because until that gap closes, a fourth plant is optionality rather than output.