Infineon has agreed to buy C2i Semiconductors, a Bangalore-based chip design startup that builds software-defined multiphase controllers and smart power stages for AI data centers. The German power-semiconductor giant announced the deal on 24 August and expects it to close in the third quarter of calendar year 2026. Financial terms were not disclosed.
The purchase is small by acquisition standards, but it points at one of the sharpest bottlenecks in the AI buildout today: getting electricity from the wall to the silicon without wasting it, melting a connector, or starving a GPU mid-training run.
Why an AI data center needs smarter power chips
A single modern AI accelerator can pull more than 1,000 watts on its own, and racks now pack dozens of them shoulder to shoulder. As those processors switch between idle and full throttle thousands of times per second, the voltage they need sags and spikes just as fast. The power delivery system has to track those swings in microseconds while keeping conversion losses low, because every wasted watt becomes heat that the cooling system must remove at additional cost.
That is where C2i's specialty comes in. Multiphase controllers orchestrate dozens of power converter stages in parallel, spreading the load so no single component overheats. Smart power stages pair the controller with the actual switching hardware and report back temperature, current, and voltage telemetry in real time. Software-defined power takes the concept further: control loops that traditionally lived in fixed analog circuits move into firmware, letting operators retune how a server draws power without touching the hardware.
C2i's work extends up the stack too. The company designs system-level architectures for vertical power delivery, including Substrate Integrated Voltage Regulators, known as SIVR. In a vertical scheme, voltage regulators sit directly beneath the processor package instead of off to the side, shortening the distance current travels from millimeters to micrometers. Shorter paths mean tighter regulation and less loss, exactly what dense AI accelerators demand.
A center of excellence in Bangalore
Adam White, president of Infineon's Power Systems division, said the deal creates "a new center of excellence for digital power technologies in India." He credited C2i's team with deep expertise in software-defined power management, backed by engineers with decades of combined industry experience.
Ram Anant, who founded C2i and remains its chief executive, said joining Infineon gives the startup access to world-class manufacturing technologies and global customer relationships. His plan, according to the joint statement: accelerate development of software-defined power architectures and ship them to customers worldwide.
The transaction also expands Infineon's engineering footprint in a country it already treats as a strategic hub. The company employs roughly 2,800 people across India, working on chip design, software, and embedded systems. Buying C2i adds a specialized digital-power team without waiting years to grow one organically.
India has spent recent years courting exactly this kind of deal. Government incentive programs have pulled in fab construction and assembly plants, but domestic design houses serving global customers remain the quieter success story. Global chip companies now run large design centers in Bangalore, Hyderabad, and Noida, and acquisitions like this one formalize their stake in the local talent pool.
The numbers behind Infineon's AI bet
Infineon is not shopping casually. Its fiscal third quarter, which ended 30 June 2026, closed with record revenue of €4.172 billion. CEO Jochen Hanebeck called power supply solutions for AI data centers the company's most important growth driver, noting very high demand that continues to outpace supply.
The company also disclosed multi-year capacity reservation agreements with leading AI customers — some signed, some still under negotiation — worth a cumulative volume in the high single-digit billions of euros. Customers are effectively prepaying to lock in power components before their data center projects break ground. For fiscal 2026 overall, Infineon expects about €16.3 billion in revenue with adjusted free cash flow around €1.85 billion.
Against a rising tide of grid investment worldwide, owning the intellectual property from wall socket to processor socket matters more than it used to. Infineon already sells silicon carbide and gallium nitride devices that convert power more efficiently than legacy silicon, plus complete power systems for hyperscale customers. C2i's controllers and stages give the company a missing layer: the intelligence that decides, moment to moment, how that conversion should behave.
A crowded field moving fast
Infineon is not alone in chasing AI power delivery. Texas Instruments, Monolithic Power Systems, Delta Electronics, and Vicor all compete in various slices of the market, and each major GPU vendor maintains its own reference designs. Vertical power delivery in particular has drawn heavy investment as packaging advances make it practical; SIVR-style approaches integrate regulators directly into the package substrate.
Buying rather than building makes sense when the technology gap is measured in product cycles. Multiphase controller design rewards accumulated know-how — layout tricks, protection schemes, telemetry calibration — that takes years to mature. C2i brings a working product line and engineering bench; Infineon brings scale, fabrication capacity, and relationships with every major server maker.
The financial logic shows in Infineon's margins. The company guided its fourth fiscal quarter to segment result margin near 23 percent, well above the 19.1 percent posted in Q3, helped along by richer mix in AI-adjacent products.
Power became the new performance ceiling
For most of computing history, chip buyers compared clock speeds and core counts. In AI data centers today, the comparison starts with kilowatts. Operators weigh how many accelerators a facility can feed before transformers saturate, breakers trip, or cooling plants fall behind. Industry roadmaps that once projected steady density gains now bend around electricity: a processor that cannot be powered cheaply enough simply does not ship at volume.
That shift explains why power components have become boardroom matters. Hyperscalers sign multi-year reservation agreements for power stages the way they once signed for GPUs themselves, because a late shipment of voltage regulators can idle an entire hall of expensive processors. Every point of conversion efficiency saved between the grid and the die compounds across hundreds of thousands of units — less heat, smaller cooling bills, more of the data center's power budget doing actual computation.
Regulators sitting closer to the processor amplify those savings. Vertical delivery architectures cut the resistance in the path from converter to compute core, which trims the voltage droop during sudden load spikes. When a training cluster jumps from standby to full tilt in milliseconds, that headroom determines whether the accelerator holds its clock or throttles. C2i's SIVR work targets precisely this problem, and Infineon's packaging expertise gives it a route into substrates that standalone controller firms cannot reach alone.
What happens next
The deal closes later this calendar year, subject to regulatory approvals. C2i's Bangalore team will operate as part of Infineon's Power Systems division, feeding both product development and the broader R&D network. For data center operators watching rack densities climb past 100 kilowatts, the practical payoff arrives as servers that squeeze more computation from every megawatt — and waste less of it as heat.
For India's chip industry, the acquisition lands as validation that local design work can command global strategic value. And for Infineon, the message to competitors is plain: the company that dominates power semiconductors intends to dominate power software too.
Sources
How AI workloads strain the power grid
Read the full announcement in the official Infineon press release, and see Economic Times coverage of the deal for details on India's role. Background on Infineon's record quarter appears in the company's Q3 FY2026 results release. Additional context on the AI power market comes via Data Center Dynamics.
