Arcium unveils Blackthorn, a confidential AI infrastructure using Nvidia GPUs for secure data handling

Here’s a question that should keep every enterprise AI team up at night: who else can see your data while it’s being processed in the cloud? The answer, for most setups, is uncomfortably long. Cloud providers, infrastructure operators, and anyone with privileged access to the hardware can, in theory, peek at data while models crunch through it.

Arcium thinks it has a fix. The Solana-based confidential computing network has announced Blackthorn, an infrastructure layer that encrypts data on Nvidia GPUs so that no single node or operator ever touches plaintext information. The pitch is straightforward: run AI inference and training at frontier scale, with privacy baked in, at speeds and costs comparable to traditional, non-confidential systems.

How Blackthorn actually works

The core technology relies on multi-party computation, or MPC. Instead of trusting one machine with your raw data, the workload gets split across multiple GPU nodes, each holding only an encrypted fragment. No single participant can reconstruct the original data.

What makes Blackthorn’s approach notable is that it runs on standard, off-the-shelf Nvidia hardware. Specifically, Arcium says it supports the Hopper H100 and H200 chips, as well as the newer Blackwell B200 series. Rather than depending on specialized trusted execution environments or custom silicon, Blackthorn achieves confidentiality through software.

The practical impact of that decision is significant. Arcium claims the software-only approach expands the pool of GPUs capable of confidential computing from roughly 2,000 units to approximately 5 million. That’s a 2,500x increase. The network effectively converts those GPUs into what Arcium describes as a collective “encrypted supercomputer,” with each participating GPU contributing compute power while maintaining cryptographic isolation of the data it processes.

The business case and who cares

The sectors most likely to pay attention are the ones where data sensitivity isn’t optional. Healthcare organizations handling patient records, financial institutions processing transaction data, and any enterprise subject to strict regulatory frameworks around data residency and privacy.

Arcium itself has been building toward this moment for a while. The project originally launched under the name Elusiv before rebranding. It secured $5.5 million in a strategic funding round in May 2024, bringing total fundraising to $9 million at that time. The project has since raised over $15 million total, with backing from Jump Crypto and Coinbase Ventures. The company was also accepted into Nvidia’s Inception Program in 2025, a startup accelerator that provides access to Nvidia’s technical resources and go-to-market support.

What this means for investors

The ARX token, which powers Arcium’s network, went through its token generation event on June 22, 2026. At launch, roughly 209 million tokens entered circulation out of a total supply of 1 billion, representing about 20.88% of the full allocation, with a market valuation of approximately $400 million. The token serves triple duty: staking, governance, and paying transaction fees within the ecosystem.

Blackthorn is still in a pre-launch phase described as “coming soon.” The technology claims are ambitious, particularly around matching the speed and cost of non-confidential AI processing. MPC protocols historically carry significant computational overhead. If Blackthorn has genuinely solved that performance penalty at frontier model scale, it would represent a meaningful breakthrough.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.