Navitas Semiconductor Corp's shares are surging in pre-market trading, becoming the latest beneficiary of the AI infrastructure investment boom. The company announced a collaboration with the NVIDIA MGX ecosystem to advance 800 VDC artificial intelligence infrastructure, boosting market sentiment.
On May 29th, the company participated in NVIDIA's partner ceremony at the Nangang Exhibition Center in Taipei, China, and is showcasing its 800 V to 6 V DC-to-DC power distribution board (PDB) product at the COMPUTEX 2026 exhibition. This product is specifically designed for 800 VDC rack architectures in AI data centers, aiming to enhance system efficiency and power density.
Navitas Semiconductor CEO Chris Allexandre stated that as AI workloads continue to expand, power delivery has become one of the most critical challenges in building next-generation gigawatt-scale AI factories.
Following the announcement, Navitas Semiconductor's stock surged over 22% in pre-market trading. The stock has already accumulated a 335% gain over the past year, with a year-to-date increase of 262%, bringing its market capitalization to approximately $6.21 billion.
Eliminating Intermediate Conversion, Directly Powering GPUs
The 800 V to 6 V power distribution board showcased by Navitas Semiconductor is a core product for its participation in the NVIDIA MGX ecosystem.
The key innovation of this product lies in directly eliminating the traditional 48 V intermediate bus converter (IBC) stage within the compute server tray, thereby maximizing system efficiency and reliability while saving valuable physical space.
From a technical specification perspective, this power distribution board is equipped with 16 GaNFast field-effect transistors, rated at 650 V with an on-resistance of 11 milliohms. It utilizes the latest DFN8×8 dual-sided cooling package, targeting a peak efficiency of 97.5%, an operating switching frequency of 1 MHz, and a power density as high as 2100 W/in³.
Furthermore, the board is approximately 20% thinner than a mobile phone. Its ultra-thin profile allows for extremely close integration with the GPU board, maximizing transient performance and improving power distribution efficiency.
Wide Bandgap Semiconductor Portfolio: Covering Full Power Chain from Grid to GPU
What Navitas Semiconductor is showcasing through this collaboration is not just a single product, but its wide bandgap (WBG) semiconductor technology portfolio designed to meet the full-chain power needs of AI factories.
In the silicon carbide (SiC) domain, the company's GeneSiC solutions support efficient power transmission from the grid to AI compute racks, covering solid-state transformers using 2300 V and 3300 V SiC power modules, as well as high-power three-phase power supply units based on fifth-generation 1200 V SiC MOSFETs.
In the gallium nitride (GaN) domain, GaNFast technology, with its MHz-frequency operation capability, higher power density, and faster transient response, supports efficient power transmission directly from the rack level to the GPU. The company currently holds over 300 granted or pending patents.
Behind the Strong Stock Performance: Revenue Exceeds Expectations, but Valuation Concerns Linger
The recent strong performance of Navitas Semiconductor's stock is partly attributable to marginal improvements in its fundamentals.
The company previously reported first-quarter fiscal 2026 revenue of $8.6 million, exceeding market expectations of $8.18 million. However, the loss per share was $0.15, significantly worse than the expected loss of $0.05.
Concurrently, the company recently completed an at-the-market equity offering raising $122 million. It entered into a sales agreement with Craig-Hallum Capital Group LLC and UBS Securities, allowing for the issuance of up to $125 million worth of Class A common stock. The proceeds are expected to support the company's strategic initiatives in high-power markets and AI infrastructure.
It is worth noting that, according to analysis, following the stock's substantial year-to-date climb, it is currently trading above estimated fair value.
Comments