Navitas Semiconductor unveils new NVIDIA-linked AI technology. Can it reverse its revenue decline?

Navitas Semiconductor Corporation (NASDAQ: NVTS) has announced a new 24-kilowatt power delivery board developed in collaboration with NVIDIA Corporation (NASDAQ: NVDA), expanding its technology portfolio for next-generation artificial intelligence data centers. Unveiled on October 12, 2026, the solution converts 800-volt direct-current power directly to 12 volts, potentially simplifying the electrical architecture used to deliver energy to high-performance computing systems. The development advances Navitas Semiconductor’s transition away from lower-power consumer applications toward AI infrastructure, electricity-grid equipment and other high-power markets. However, the commercial opportunity remains dependent on customer adoption, manufacturing scale and the company’s ability to convert technology demonstrations into profitable product shipments.

The new board uses Navitas Semiconductor’s gallium nitride technology and is designed for integration into NVIDIA’s MGX ecosystem. It is scheduled to be showcased in the NVIDIA AI Factory MGX partner display area at the 2026 Open Compute Project Global Summit.

For Navitas Semiconductor, the announcement represents another step in its effort to position power electronics as an essential component of expanding AI infrastructure investment. The company has already demonstrated a related 20-kilowatt solution designed to convert 800 volts to six volts.

How does Navitas Semiconductor’s 24 kW power board address the electricity demands of NVIDIA AI systems?

Electricity delivery has become a growing engineering consideration for artificial intelligence infrastructure as individual computing systems accommodate increasingly powerful processors and greater computing density. Traditional electrical architectures use several conversion stages to transform incoming power into the lower voltages required by processors and associated components. Each conversion stage consumes space, introduces additional components and creates some energy losses.

Navitas Semiconductor’s latest design uses a single-stage architecture to convert 800-volt direct current directly to 12 volts. The company says the approach eliminates an intermediate 50-volt-to-12-volt conversion stage and simplifies the power path feeding voltage regulator modules near computing processors. Reducing the number of conversion stages could potentially improve efficiency while allowing more compact equipment designs.

The board is designed to deliver 24 kilowatts with a targeted full-load efficiency of 97% and power density exceeding 2,000 watts per cubic inch. These specifications describe the company’s intended performance characteristics and should not be interpreted as independent confirmation of performance in commercial data-center operations.

At a 24-kilowatt output and 97% conversion efficiency, the corresponding input would be approximately 24.74 kilowatts, implying conversion losses of roughly 0.74 kilowatts at that operating point. This illustrative calculation helps explain why efficiency improvements can become financially relevant in high-density infrastructure. Actual electricity savings would depend on the efficiency of the complete power system, including the conversion stages displaced and the equipment operating conditions.

The board has been designed for a 400-millimetre by 60-millimetre form factor and uses gallium nitride power devices. Gallium nitride is attractive for certain high-frequency power conversion applications because its electrical properties can help enable compact switching systems. However, overall performance also depends on thermal management, packaging, control electronics and the requirements of the customer’s architecture.

Why is the NVIDIA collaboration strategically important for Navitas Semiconductor’s AI infrastructure business?

NVIDIA has become a central participant in the development of AI computing platforms, creating a broad ecosystem of companies supplying supporting infrastructure. Power equipment manufacturers are competing to contribute technologies capable of supporting increasingly demanding computing architectures. Navitas Semiconductor’s collaboration places its technology within that development environment.

The company has been shifting its business toward high-power applications through its Navitas 2.0 strategy. Its portfolio includes gallium nitride and silicon carbide technologies for data centers, energy infrastructure and industrial power systems. The strategy seeks to replace reliance on lower-power consumer electronics with applications where customers require increasingly sophisticated power delivery and conversion.

The October announcement builds on earlier collaboration involving an 800-volt-to-six-volt board demonstrated during COMPUTEX 2026. The development of multiple conversion architectures gives potential customers alternatives suited to different processor, voltage-regulation and system-design requirements. These product variations may broaden Navitas Semiconductor’s opportunities without ensuring that any specific platform will achieve large-scale commercial deployment.

The NVIDIA association has clear marketing and technical relevance, but it is important to distinguish product collaboration from contracted revenue. Navitas Semiconductor has not disclosed a purchase commitment or future revenue value attached to the newly announced 24-kilowatt board. Participation in the MGX ecosystem does not independently establish that NVIDIA or its customers will procure the technology in material volumes.

Commercial adoption will depend on successful qualification by system manufacturers, competitiveness against alternative power technologies and the timing of new AI infrastructure deployments. Navitas Semiconductor must also demonstrate that it can produce components reliably at the volumes and costs required by large infrastructure customers.

What do Navitas Semiconductor’s latest revenue and cash figures reveal about its turnaround?

Navitas Semiconductor reported second-quarter 2026 revenue of approximately $10.5 million, an increase of 22% from the preceding quarter’s $8.6 million. The improvement indicates that sequential growth was returning as the company moved toward higher-power applications. However, revenue remained below the $14.5 million reported in the second quarter of 2025, demonstrating that the business was still transitioning away from its previous revenue mix.

The company’s GAAP operating loss was approximately $27.2 million during the quarter, compared with $27.8 million in the preceding quarter. Its non-GAAP operating loss was approximately $11.4 million, while non-GAAP gross margin reached 39.5%. The difference between GAAP and adjusted results reflects accounting adjustments, including acquisition-related intangible amortisation, and both measures are relevant when evaluating the financial position.

Navitas Semiconductor’s GAAP net loss of approximately $228.2 million included a substantial non-cash charge associated with the final remeasurement of earnout liabilities. Consequently, that reported loss should not be confused with an equivalent quarterly operating cash outflow. Nevertheless, recurring operating losses remain a central consideration as the company invests in product development and sales expansion.

The balance sheet provides financial flexibility. Navitas Semiconductor ended June 2026 with approximately $557.4 million in cash and cash equivalents, compared with $236.9 million at the end of 2025. That liquidity can support product qualification, operating expenditure and strategic investment, although it does not remove the need for sustainable commercial economics.

In July, management projected third-quarter revenue between $13 million and $14 million, representing approximately 28% sequential growth at the midpoint. It also expected full-year 2026 revenue to increase by a mid-single-digit percentage compared with the prior year, despite the withdrawal from substantial parts of its historical mobile-device business.

Management anticipated that AI infrastructure, including data centers and grid-related applications, would contribute more than one-third of total revenue by the end of 2026. That is a forward-looking sales-mix expectation rather than a completed financial milestone.

Why did Navitas Semiconductor shares decline even as AI infrastructure opportunities expanded?

Navitas Semiconductor shares closed at $9.99 on October 12, down approximately 7.3% from the previous session. Importantly, the new NVIDIA collaboration announcement was released after the regular United States trading session closed, so the day’s decline should not be described as a reaction to the newly unveiled board. The closing price instead reflects trading that occurred before that announcement became public.

The stock has experienced considerable volatility as investors assess whether AI infrastructure demand can create a durable revenue recovery. The company’s technology announcements provide indications of product development, while its financial results show that the revenue base remains relatively small compared with the scale of the opportunity being pursued.

The investment case therefore involves two separate developments. The first is the transition toward high-power semiconductor applications, where AI infrastructure and energy systems could support higher product demand. The second is the pace at which those opportunities become customer orders, shipments and recurring revenue.

Competition is another important consideration. The electrical architecture of AI data centers attracts established power management companies, semiconductor manufacturers and specialised component suppliers. Customers will evaluate efficiency, reliability, thermal performance, cost and long-term supply availability rather than select components solely because they participate in the NVIDIA ecosystem.

Navitas Semiconductor has a stronger cash position and a clearer strategic focus than a simple review of its recent annual revenue decline might suggest. The newest power delivery board also broadens the technological capabilities available to potential AI infrastructure customers. However, its commercial significance will become measurable only when the company reports production qualifications, customer ramps and associated sales.

The next quarterly results should provide evidence of whether the high-power transition is progressing at the pace management previously anticipated. Revenue growth toward the July guidance range, a larger contribution from AI infrastructure and continued margin improvement would support the operational case for Navitas 2.0. Weak customer conversion or persistent operating losses would make the company’s technological achievements harder to translate into a convincing financial recovery.

What are the key takeaways from the Navitas Semiconductor and NVIDIA power board announcement?

• Navitas Semiconductor unveiled a new 24-kilowatt AI power delivery board.
• The design was developed in collaboration with NVIDIA.
• It converts 800 volts directly to 12 volts.
• The targeted full-load efficiency is 97%.
• The technology is planned for display at the 2026 Open Compute Project Global Summit.
• No material purchase commitment or contract value was disclosed.
• Second-quarter revenue reached $10.5 million, up sequentially but below the prior-year quarter.
• Cash and cash equivalents stood at approximately $557.4 million at June 30.
• The October 12 stock decline occurred before the product announcement was released.
• Future results must show whether AI infrastructure adoption supports sustained revenue and margin growth.