Summary

  • Enphase says its Austin test system converted 13.8 kV AC to a stable 800 VDC output through simulated AI-rack load changes. It consisted of 132 power modules, with 44 in series on each of three phases.
  • Third-quarter 2026 revenue was $293.3m, down 28.5% from $410.4m a year earlier. The company plans customer pilots in 2027 and expects commercial IQ SST shipments to start in 2028.

A convincing bench test, an unforgiving calendar

In the technical announcement issued on 27 October, Enphase describes a serious piece of electrical engineering. Its IQ solid-state transformer converted medium-voltage alternating current to an 800-volt direct-current output in an Austin research lab. The output held steady during load steps intended to imitate the changing appetite of AI computing racks. The company connected 132 modules as one system, 44 in series on each of three phases. Each module is rated at 5.5 kW; distributed controls kept voltage sharing balanced across them.

The valuable result is specific. A bank of modules handled a 13.8 kV input and kept its output stable through a defined test. Enphase's next stated milestones are 34.5 kV operation and a full rack working at megawatt scale by the end of 2026. That second milestone is still on the company's to-do list. So are customer pilots in 2027. Commercial shipments are expected to begin in 2028. One can admire the laboratory result and still put those dates on different lines of the investment model.

The quarterly results published on the same day show why the interval matters. Revenue of $293.3m was almost flat against the preceding quarter's $291.9m but 28.5% below the $410.4m recorded a year earlier. GAAP gross margin fell to 43.3% from 47.8% a year before. GAAP operating income shrank to $6.5m from $66.2m. Enphase generated only about $0.1m of operating cash during the quarter while spending $11.5m on capital equipment. It closed September with $931.8m of cash, equivalents and marketable securities. That is a substantial cushion; it is also the pot from which the company must help fund a product with a commercial clock two years away.

The bridge cannot be paid in volts

The old business is not standing still while the new one waits. Enphase shipped approximately 1.42m microinverters and 129.9 MWh of batteries in the quarter. Its Q4 revenue guide of $225m to $255m sits below Q3's actual $293.3m throughout the range. The company expects global sell-through to fall 10% sequentially in Q4. Its forecast includes $51.2m of safe-harbor revenue, after $85m in Q3. Those figures leave a current sales engine facing another weak quarter just as management asks the market to value a future data-centre engine.

The company says its IQ SST customer and partner pipeline has more than doubled to over 50. It describes work with a leading hyperscaler and a global AI-computing leader without naming them. A pipeline counts conversations and prospects; the release gives no signed order value. The more useful economic sequence runs through qualification, pilot operation, procurement and shipment. A data-centre buyer controls the acceptance step. Enphase controls the architecture, manufacturing preparation and capital it commits before that acceptance. Neither side can skip the calendar by calling a test a sale.

Enphase proposes to remove conversion stages between medium-voltage supply and an 800 VDC computing rack. Fewer stages could create a valuable power and space proposition, but the releases publish no measured system-wide efficiency or field operating record. The company has made a sharp engineering claim and given the market dated checkpoints. The next proof is a megawatt-scale rack that runs at rated power, followed by a customer pilot with disclosed operating results and a commercial order. Until those appear, the loudest financial fact remains the revenue decline that has to carry the development period.

Sources