3 hours ago
1
A Bitcoin mining company just pulled off something that should make Nvidia’s pricing team slightly uncomfortable. HIVE Digital Technologies trained large language models on older A40 GPUs in Paraguay and achieved performance that matched Nvidia’s flagship H100 chips, the same hardware that companies are currently paying a premium to get their hands on.
The trick wasn’t better hardware. It was smarter software. And the market noticed, with HIVE’s stock climbing as much as 22% on the news.
What HIVE actually proved
HIVE partnered with Columbia University’s Department of Industrial Engineering and Operations Research to run its first external academic AI project. The setup: Columbia researchers in New York remotely operated HIVE’s GPU infrastructure in Asunción, Paraguay, roughly 5,000 miles away.
Over approximately two months, the team trained large language models with up to 1.4 billion parameters on HIVE’s Nvidia A40 GPUs. After optimization and normalization, the A40 systems matched the performance of Nvidia’s far more expensive H100 chips for those workloads.
The results were compelling enough that the team submitted their findings to NeurIPS, one of the most prestigious machine learning conferences in the world.
Why older GPUs performing like newer ones matters
There’s an important caveat here. The performance parity was demonstrated at up to 1.4 billion parameters. The largest frontier models today run into the hundreds of billions or even trillions of parameters. Whether A40s can keep pace with H100s at those scales is a very different question, and almost certainly the answer is no.
The intercontinental angle matters too. Columbia researchers operated the Paraguay facility remotely from New York across 5,000 miles. That’s a proof-of-concept for distributed AI training workflows, where compute happens wherever electricity is cheapest and researchers work wherever they happen to be.
HIVE’s broader play
HIVE Digital Technologies, listed on both the TSX and Nasdaq under the ticker HIVE, has been gradually repositioning itself from a pure-play crypto miner into a high-performance computing and AI data center operator. The Paraguay facility is central to that strategy.
Paraguay offers something that most AI data center markets don’t: abundant, cheap, renewable hydroelectric power.
The stock market response reflected that distinction. HIVE shares gained between 15.5% and 22% following the announcement.
What this means for investors
The immediate takeaway is that HIVE has established baseline performance data for its GPU infrastructure. That data now underpins the company’s expansion plans, giving management something concrete to point to when pitching future AI compute deals to potential customers.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) · 