AI power hopes, factory risk attached
- Navitas is pivoting from phone chargers to high-power markets like AI data centers, grid equipment, and industrial systems.
- Q1 2026 revenue rose 18% from Q4 to $8.6 million, but it was still down 39% from the prior year.
- AI infrastructure, meaning data center and grid sales, grew 50% quarter over quarter in Q1 2026.
- The balance sheet had $221 million in cash at the end of Q1 2026, but the company is still losing money.
- The biggest hard risk is supply: TSMC plans to stop GaN production in July 2027, so Navitas must qualify new suppliers.
A cleaner pivot, not a clean win
Navitas is trying to become a power-chip supplier for the AI and energy buildout. Its chips help convert electricity more efficiently. That matters when AI racks, solar systems, and grid gear need more power in less space.
The bull case got better in Q1 2026. Revenue returned to sequential growth, up 18% from Q4 to $8.6 million. Management said high-power revenue grew 25% year over year, and AI infrastructure grew 50% quarter over quarter.
The bear case is still real. Revenue was down 39% from the prior year because mobile and consumer demand, especially in Asia and China, fell hard. The company also posted a $27.8 million operating loss in Q1 2026, far larger than its quarterly sales.
This is a high-risk transition story. The cash balance gives Navitas time, with $221 million in cash and no debt in the internal thesis. But the company needs AI data center and grid programs to turn into binding orders before the old mobile business fades away.
Fabless chips, fragile supply
Navitas is a fabless semiconductor company. That means it designs chips, but outside partners make the wafers, assemble parts, and test them. This keeps capital spending lower, but it makes factory access, yields, and supplier timing critical.
The company sells power semiconductors through distributors to equipment makers and their suppliers. Its older revenue came from mobile and consumer chargers. Navitas 2.0 shifts the sales push toward AI data centers, energy and grid infrastructure, performance computing, and industrial electrification.
A key change is the GaN intellectual property cross-license with Infineon. In plain English, this can help large customers get a second source for similar GaN parts, which lowers the fear of relying on one small supplier.
The model breaks if high-power customers stay in pilot mode, or if new suppliers cannot replace TSMC for GaN before July 2027. It also breaks if SiC pricing in China is hurt by tariffs for too long.
Power chips for bigger loads
GaN power ICs
Gallium nitride, or GaN, chips switch power very fast and can shrink power supplies. Navitas is aiming these parts at AI racks, high-power computing, and energy systems.
High-voltage SiC devices
Silicon carbide, or SiC, handles high voltage and heat better than older silicon in many uses. Navitas targets high-power markets, but tariffs on US-sourced SiC wafers are a major pricing risk in China.
Bi-directional GaN platforms
These products can move power in both directions, which helps in solar and energy storage systems. They fit the grid and energy side of the Navitas 2.0 plan.
Low-voltage AI rack platforms
Navitas has 80V to 120V platforms aimed at power delivery inside AI data center racks. The upside depends on hyperscalers raising rack power density and adopting newer power designs.
Legacy mobile and consumer chargers
This was the old core business, but it is being wound down. Management said mobile contribution should become insignificant by year-end 2026.
Q1 sales by region
This mix is from Q1 2026 revenue by end-customer country in the latest 10-Q. It is geographic revenue, not end-market revenue, so it does not show the exact split between AI, grid, mobile, and industrial uses.
What could break the story
GaN supplier switch
High impact · High oddsNavitas relies on TSMC as its sole GaN wafer supplier in the internal thesis. TSMC announced it plans to stop GaN production in July 2027. Navitas is working with alternatives such as Powerchip, but new suppliers must meet cost, quality, volume, and timing needs.
AI programs stay non-binding
High impact · Medium oddsNvidia named Navitas as a partner in an AI data center power initiative, which is good validation. But a partnership mention is not the same as a purchase order. The stock needs real revenue from 800V HVDC and other AI power designs, not only press releases.
Old mobile revenue fades too fast
High impact · Medium oddsMobile and consumer sales are shrinking by design. Q1 2026 revenue was down 39% year over year, mainly from lower mobile sales in Asia, especially China. If high-power sales do not scale quickly, total revenue can stay too small to cover costs.
Losses burn cash
Medium impact · High oddsNavitas had $8.6 million of Q1 2026 revenue and a $27.8 million operating loss. Operating cash use was $16.4 million in Q1 2026. The $221 million cash balance helps, but long losses can still force more share sales or spending cuts.
China and tariff pressure
Medium impact · Medium oddsChina was 36% of Q1 2026 revenue, down from 41% a year earlier. The company has also warned that 125% tariffs on US-made chips could hurt SiC competitiveness in China because it relies on a single US source for SiC wafers. That can slow sales or pressure pricing.