The global race to break Nvidia’s near-monopoly on high-end artificial intelligence silicon has taken a major leap forward. Dutch AI chipmaker Euclyd has successfully secured a massive $231 million (€200 million) Series A funding round. This significant capital injection was co-led by consumer electronics giant Samsung, alongside Somerset Capital Partners, Innovation Industries, and the EQT-managed Scaleup Europe Fund. Launched in 2024, Euclyd is designing specialized chip systems for AI inference, utilizing an alternative processor and memory architecture designed to bypass traditional graphics processing units (GPUs).
While the technological implications of this deal are clear, the financial and regulatory undercurrents are equally profound. As tech hyperscalers like Google, AWS, Meta, and OpenAI (with its newly announced Jalapeño chip) race to develop custom silicon, the underlying fiscal frameworks governing global trade, intellectual property (IP) licensing, and Indirect Taxes (GST) are facing unprecedented disruption. This shift in the hardware paradigm is not just a technological pivot; it is a complex cross-border compliance challenge.
The Shift in AI Infrastructure and Revenue Models
According to Euclyd’s CEO Bernardo Kastrup, the true potential of artificial intelligence will remain constrained unless the global technology sector fundamentally overhauls the infrastructure beneath it. Traditional GPUs, originally designed for gaming, have struggled under the soaring energy demands and astronomical operational costs of modern AI data centers. Euclyd aims to solve this by rolling out physical chip systems by 2028, with a target of serving thousands of enterprise customers by 2030.
To achieve this, Euclyd is targeting two distinct revenue channels:
- Selling physical server racks to enterprises for secure, on-premise AI inference workloads.
- Licensing its proprietary chip designs to enterprises and hyperscalers seeking to manufacture custom silicon.
While this dual-model strategy maximizes market reach, it introduces severe tax complexities, particularly concerning cross-border transactions, Goods and Services Tax (GST) classification, and transfer pricing.
Decoding the GST and Cross-Border Tax Implications
The transition of AI infrastructure from a centralized cloud-based model to localized, on-premise hardware and custom-licensed silicon triggers a cascading series of tax implications. These challenges are highly relevant as companies navigate the fiscal aftershocks of tech layoffs and AI restructuring globally.
1. The Dichotomy of Goods vs. Services under GST
Euclyd’s dual-revenue model creates a classic tax classification dilemma. Selling physical server racks constitutes a “supply of goods.” In contrast, licensing chip designs is categorized as a “supply of services” (specifically, the transfer of Intellectual Property Rights). Under GST regimes worldwide, including India’s GST framework, these two streams are treated entirely differently:
- Physical Server Racks: Importing hardware components and fully assembled server racks involves customs duties and Integrated GST (IGST). Companies must meticulously manage their Input Tax Credit (ITC) to ensure that taxes paid on imported hardware components are successfully offset against domestic sales liabilities.
- IP Licensing: Licensing chip architecture is subject to GST under the Reverse Charge Mechanism (RCM) when imported by domestic entities. If a domestic enterprise licenses Euclyd’s designs, they must self-assess and pay GST on the import of services, which can severely impact short-term cash flows if ITC utilization is delayed.
2. Transfer Pricing and Strategic Partnerships
Samsung’s involvement in Euclyd extends far beyond mere financial backing. As Kastrup noted, Samsung’s expertise in memory manufacturing, system engineering, and global supply chains is a vital asset. However, when a major investor also acts as a supply chain partner, tax authorities scrutinize the transactions under Transfer Pricing (TP) regulations.
Any transactions between Euclyd and Samsung’s manufacturing arms—such as the procurement of high-bandwidth memory chips or joint engineering initiatives—must be conducted at Arm’s Length Pricing (ALP). If tax authorities deem that the pricing of these components or services has been manipulated to shift profits to lower-tax jurisdictions, both entities could face severe transfer pricing adjustments, penalties, and protracted litigation. This mirrors the complex tax landscapes seen in other major tech ecosystems, similar to how Nvidia’s talent spin-offs and cross-border tax compliance require rigorous regulatory alignment.
3. Double Taxation and Withholding Taxes (WHT) on Royalty Payments
Cross-border licensing of proprietary chip designs inevitably triggers withholding tax obligations. When an enterprise in one country pays licensing fees or royalties to Euclyd in the Netherlands, that payment is typically subject to withholding tax in the source country. Businesses must carefully navigate Double Taxation Avoidance Agreements (DTAAs) to mitigate these tax leakages. Failure to establish proper documentation, such as Tax Residency Certificates (TRC), can lead to higher tax withholding rates, directly diminishing the net revenue realized by the startup.
4. Supply Chain Relocation and Customs Valuation
By partnering with Samsung, Euclyd gains access to a massive global supply chain. However, manufacturing physical chip systems and shipping them across multiple international borders introduces customs valuation risks. Customs authorities frequently scrutinize the relationship between buyers and sellers of high-tech components. If the transaction value of imported chip systems is suspected of being undervalued to minimize customs duties and import GST, it can lead to shipment delays, audits, and heavy fines.
The Strategic Path Forward for AI Hardware Innovators
As the AI hardware landscape diversifies, startups and tech conglomerates cannot afford to treat tax compliance as an afterthought. Designing revolutionary architecture is only half the battle; the other half is structuring the business in a tax-efficient, compliant manner. Companies must proactively address:
- Establishing robust transfer pricing documentation for all joint ventures and co-development agreements.
- Clearly segregating hardware sales from IP licensing contracts to avoid classification disputes with GST authorities.
- Optimizing supply chains to leverage favorable trade agreements and minimize import duty exposure.
Ultimately, the success of Nvidia’s emerging rivals will depend not just on their engineering prowess, but on their ability to navigate the complex, shifting maze of international tax law and GST compliance.
Frequently Asked Questions
The €200 million ($231 million) Series A funding round was co-led by Samsung, Somerset Capital Partners, Innovation Industries, and the EQT-managed Scaleup Europe Fund.
Euclyd is developing specialized chip systems for AI inference that utilize an alternative processor and memory architecture distinct from traditional graphics processing units (GPUs).
Euclyd is targeting two revenue channels: selling server racks to enterprises for secure, on-premise AI inference, and licensing its custom chip designs to companies developing custom silicon.
Euclyd aims to begin rolling out its physical chip systems in 2028, with the goal of serving thousands of enterprise customers by 2030.
