Navitas Agrees To Pay Up To $232.8 Million For AI Power Startup Claros

Navitas Semiconductor agreed to pay up to $232.8 million for Claros, a startup developing vertical power delivery technology for AI data centers, with $216 million in cash and stock at closing and up to $16.8 million in additional stock tied to milestones. The acquisition, announced August 24 and expected to close before the end of 2026, aims to address power delivery challenges for AI chips like Nvidia's GB200 modules, which can draw around 2,700 watts. Navitas said the deal would more than double its identified 2030 serviceable addressable market to over $8 billion.

Navitas Semiconductor is paying up to $232.8 million for Claros because AI chips now have a power problem sitting right under the package. If you follow the AI hardware race only by watching GPUs, you're missing the wiring that decides how far those GPUs can go. Navitas announced the Claros deal on August 24, and the price tells you what has changed in AI infrastructure. This is not a small tuck-in for a charger company. Navitas said in its GlobeNewswire release that it will pay about $216 million at closing in cash and Class A stock, with up to $16.8 million more in stock over the following two years if Claros hits business milestones. Both boards approved the deal, and Navitas expects it to close before the end of 2026, subject to regulatory approvals and other usual conditions. There is another stock bill attached to it. Navitas said certain Claros employees will be eligible for about $28.9 million in performance-based share compensation tied to the same milestones. The company did not disclose the cash-versus-stock split of the $216 million closing payment. That detail matters for shareholders, because a deal paid partly in stock can still be sensible while making the ownership math less tidy. What Claros makes is the real story. The company, founded in 2024 and backed by Red Cell Partners, General Catalyst, Systemiq Capital, VIPC and other investors, develops vertical power delivery and integrated voltage regulator technology for AI data centers. Navitas says Claros' approach places power conversion directly beneath or inside the chip package or printed circuit board, shrinking the trip from inches to millimeters. That's the point. Sam Altman Fears a Handful of Companies Will Control AI https://startupfortune.com/sam-altman-fears-a-handful-of-companies-will-control-ai/ OpenAI CEO Sam Altman told podcaster David Senra that his biggest fear is a small number of companies gatekeeping AI, framing it as a choice between "AI authoritarianism or liberty." The remarks land awkwardly given OpenAI's own dependence on Microsoft's cloud and Nvidia's chips, and its pending push for a $1 trillion IPO valuation. - sam altman fears ai companies controlling market access https://startupfortune.com/sam-altman-fears-a-handful-of-companies-will-control-ai/ - how many companies will dominate artificial intelligence industry https://startupfortune.com/sam-altman-fears-a-handful-of-companies-will-control-ai/ Every bit of distance between a voltage regulator and a GPU or custom AI accelerator turns into resistance, heat and lost efficiency. On older server boards, that was annoying. On racks filled with power-hungry AI processors, it becomes a hard limit. Nvidia's Blackwell generation made the issue plain: AnandTech noted that GB200 modules combine two GPUs and a Grace CPU at a total module power budget around 2,700 watts, with each GPU estimated at roughly 1,200 watts in that configuration. The fight is under the chip Navitas has spent years selling itself around GaNFast gallium nitride and GeneSiC silicon carbide power semiconductors. Claros gives it a claim at the last and most awkward step of the chain: getting current into the processor core itself. Navitas calls the broader pitch grid-to-xPU, meaning everything from higher-voltage data center distribution down to the processor socket. According to Navitas, the acquisition would more than double its identified 2030 serviceable addressable market to over $8 billion. At least $3.5 billion of that comes from vertical power delivery and integrated voltage regulator markets: its existing GaN and high-voltage or ultra-high-voltage silicon carbide business makes up another $3.5 billion, and new junction field-effect transistor technology adds roughly $1 billion on top. Do not treat that addressable market slide as revenue. It isn't. It is management's map of where it wants to compete, and maps are cheaper than design wins. But the target itself is believable because the power stack around AI servers is being rebuilt in public. Navitas showed an 800 V-to-6 V power delivery board at Nvidia GTC 2026 in March, and said the board was designed to remove the traditional 48 V intermediate bus converter stage while targeting 97.5% peak efficiency. That is not marketing fluff when data centers are fighting for watts rack by rack. SemiAnalysis has been writing about this power delivery contest for years, including the role of Vicor, Monolithic Power Systems, Delta, Analog Devices, Renesas and Infineon around AI accelerators. Vicor, for example, markets factorized vertical power delivery and current multiplier modules that sit close to or beneath processors. Monolithic Power Systems says its products power every AI socket. These are not side players waiting for Navitas to arrive. Navitas is buying into a crowded room. The hard part starts after signing Investors liked the story at first. Investing.com reported that Navitas shares rose 5.0% in pre-open trading on August 25 after the company disclosed the Claros agreement, while StockTitan's delayed market data later showed a gain above 7% during the early reaction. For a stock that has had a rough year, a visible AI infrastructure deal was always going to get attention. TCS Will Buy Porsche's MHP Consulting Unit for $373 Million https://startupfortune.com/tcs-will-buy-porsches-mhp-consulting-unit-for-373-million/ Tata Consultancy Services is buying Porsche's German consulting arm, MHP, for $373 million as part of a five-year, $1.45 billion partnership covering AI, engineering and software work. Announced August 24, the deal deepens a pattern of Indian IT giants buying up European automotive engineering talent, following Infosys, HCLTech and Persistent... - TCS acquires Porsche consulting unit MHP for millions https://startupfortune.com/tcs-will-buy-porsches-mhp-consulting-unit-for-373-million/ - Porsche outsources AI and software work to TCS https://startupfortune.com/tcs-will-buy-porsches-mhp-consulting-unit-for-373-million/ Frankly, the approval from traders is the easy part. The deal still has to close. Navitas hasn't said where Claros' products will get manufactured or how the engineering team gets folded in - and it hasn't named which customers are far enough along to turn the technology into revenue after 2028 or 2029. Navitas said the Claros technology should accelerate growth in that period while keeping its current path to profitability unchanged. That is a useful promise. It is also one the company will have to prove with actual sockets. Winning inside a Meta, Microsoft, Google or Nvidia-linked server design takes more than a clever package. Qualification cycles are long. Power parts are not decorative. If they fail, the expensive chip they feed becomes useless, so buyers move carefully even when the need is obvious. Still, the logic of the acquisition is clear. AI companies can keep asking for more compute. But every extra watt has to move through copper and converters, then get pulled off through heat sinks, before it becomes useful work. Navitas is betting that the least glamorous part of the AI stack, moving electricity a few millimeters without wasting it, is now worth nine figures. Also read: Jack Ma Buys $77 Million of Alibaba Shares After Its AI Fundraise Sent Stock Lower https://startupfortune.com/jack-ma-buys-77-million-of-alibaba-shares-after-its-ai-fundraise-sent-stock-lower/ • AliExpress Ran a Hidden Script That Kept Breaking Bluetooth Headphones https://startupfortune.com/aliexpress-ran-a-hidden-script-that-kept-breaking-bluetooth-headphones/ • Google News Keeps Pointing Readers to Stories They Cannot Actually Read https://startupfortune.com/google-news-keeps-pointing-readers-to-stories-they-cannot-actually-read/