Connect with us
DAPA Banner

Crypto World

Hive Digital reaches AI cloud milestone in Paraguay

Published

on

Crypto Breaking News

Hive Digital Technologies announces an AI cloud milestone in Paraguay, describing the development as a step in expanding its cloud capabilities for AI and research. The notice also references Columbia University LLM research spanning New York to Asuncion, indicating the initiative may involve cross-border collaboration or access to regional compute resources. While the full details are not included here, the announcement highlights a shift in the company’s cloud footprint in an important Latin American market. Readers should watch for additional disclosures on scope, participants, and practical implications as the company provides more specifics.

Key points

  • AI cloud milestone reached in Paraguay, as described by the company.
  • Release references Columbia University LLM research connecting New York and Asuncion.
  • Statement implies cross-border compute resources linked to AI research.

Why it matters

The milestone matters because it signals progress in cloud-enabled AI and potential regional access for researchers and institutions. If the initiative enables cross-border collaboration with Columbia University, it could influence how researchers plan experiments, require infrastructure, and coordinate efforts across North and South America. The public details are limited, but the move may shape attention on AI cloud deployments in Latin America and may guide investors and developers monitoring compute availability and academic partnerships in the region.

What to watch

  • Clarification of the milestone’s scope, including services and capacity.
  • Any partners or institutions involved beyond Columbia University.
  • Upcoming disclosures or timelines for broader availability.

Disclosure: The content below is a press release provided by the company or its PR representative. It is published for informational purposes.

HIVE Digital Technologies Reaches AI Cloud Milestone in Paraguay, Powers Columbia University LLM Research from New York to Asunción

This news release constitutes a “designated news release” for the purposes of the Company’s prospectus supplement dated November 25, 2025 to its short form base shelf prospectus dated October 31, 2025.

San Antonio, Texas, March 18, 2026 HIVE Digital Technologies Ltd. (TSX.V: HIVE) (Nasdaq: HIVE) (FSE: YO0) (BVC: HIVECO) (the “Company” or “HIVE”), a global leader in sustainable digital infrastructure and AI compute, today announced that its BUZZ AI Cloud platform in Asunción, Paraguay is now operational with live GPU compute nodes serving workloads on the platform by an academic research team from Columbia University in New York.

The Asunción deployment is the first GPU cluster to go live under HIVE’s phased strategy to layer AI and high-performance computing (“HPC”) infrastructure onto its existing renewable energy footprint in Paraguay. The cluster is hosted within a Tier-III data center operated by Paraguay’s largest telecommunications provider and is purpose-built to handle AI model training, inference, and computationally intensive research workloads.

Advertisement

HIVE expects to use the results of the cluster to establish the proof of concept for AI compute between New York and Asuncion. From this proof-of-concept, the Company expects to build future Tier III data center capacity in Yguazú, with infrastructure upgrades required to provide high-availability, low-latency GPU AI cloud compute from Paraguay. Paraguay’s hydroelectric generation capacity and the telecom partner’s nationwide fiber backbone provide the energy and connectivity foundation to support that growth. As regional South American institutional and commercial demand for HPC and AI Cloud develops, the pace and scale of the Company’s Tier III expansion in Paraguay will be guided by customer adoption and the Company’s capital position.

Columbia University Research Team Goes Live on BUZZ Cloud

The Columbia University team is using BUZZ Cloud GPU infrastructure to conduct research focused on large language model (“LLM”) pre-training, including end-to-end training of foundation models. The research team’s use of BUZZ Cloud infrastructure is a non-commercial research engagement intended to generate performance data that will inform the Company’s roadmap for scaling commercial HPC capacity in Paraguay. The team is developing optimization algorithms that improve model quality while reducing computational and memory costs, evaluating their methods using standard training metrics such as loss and perplexity, as well as downstream benchmarks.

Their work begins with small- to medium-scale models (0.2B to 2B parameters, including GPT-2-class and LLaMA-style architectures) and is scaling to larger models (8B+ parameters) using multi-GPU distributed training frameworks. The team’s recent focus includes improving and understanding Muon and MuonClip, the latter of which has been used in training industry-level LLMs such as Kimi K2. In early experiments, Muon has shown roughly 1.3x greater efficiency¹ than standard baselines by exploiting the structure of model weights. The team’s LLM reasoning research was recently accepted for publication by Transactions on Machine Learning Research (“TMLR”), a peer-reviewed journal hosted by the Journal of Machine Learning Research (“JMLR”).

Having the Columbia University research team run active LLM training jobs from New York on GPU infrastructure in Asunción provides HIVE with real-world performance data across latency, throughput, and workload management. The Company intends to use these findings to shape its roadmap for scaling HPC capacity in Paraguay, with initial deployment targets through 2027.

Advertisement

Paraguay: The Western Hemisphere’s Next Potential AI Infrastructure Frontier

Paraguay’s President Santiago Peña earned his Master’s degree in Public Administration from Columbia’s School of International and Public Affairs (“SIPA”) in 2003, creating a notable link between the institution whose researchers are now training LLMs on BUZZ Cloud and the nation whose clean energy powers it.

Management believes large-scale AI compute requires two resources Paraguay can deliver in abundance: reliable, low-cost electricity and fiber connectivity with the bandwidth and security to move data across long distances without degradation. HIVE’s existing 300-megawatt (“MW”) renewable power base, sourced from hydroelectric generation, combined with the telecom partner’s enterprise-grade network infrastructure, creates a platform that can serve demanding workloads originating outside Paraguay’s borders, including from North American institutional clients.

Paraguay’s economy has posted strong growth in recent quarters, backed by stable governance and a policy environment that has welcomed foreign infrastructure investment. HIVE believes those conditions, paired with the country’s distinctive energy profile and expanding digital connectivity, position the country to play a growing role in South America’s AI and high-performance computing future.

Strategic Outlook from HIVE Leadership

Frank Holmes, Executive Chairman of HIVE, stated, “HIVE has 300 MW of renewable hydroelectric power operational in Paraguay, with another 100 MW in development. Before scaling an AI factory, it’s prudent to beta test. This deployment marks our first live GPU compute workload in Asuncion and provides the real-world performance data we need to guide our Tier-III expansion roadmap. We started in Paraguay with Bitcoin mining. Layering AI and HPC infrastructure onto that existing energy base is the next phase, and this cluster is the first step in validating that approach.”

Advertisement

Aydin Kilic, President and CEO of HIVE, added, “We are taking a meaningful and impactful approach to developing a solution to being a leader of GPU AI compute and HPC in South America. Having a research team from Columbia University running LLM training workloads on HIVE’s BUZZ Cloud infrastructure in Asunción is a powerful validation of what we are building. We will use this data to validate our proof of concept for GPU Cloud AI compute from New York to Asunción and to build our roadmap for large-scale HPC capacity in Paraguay by 2027.”

About HIVE Digital Technologies Ltd.

Founded in 2017, HIVE Digital Technologies Ltd. is the first publicly listed company to mine digital assets powered by green energy. Today, HIVE builds and operates next-generation Tier-I and Tier-III data centers across Canada, Sweden, and Paraguay, serving both Bitcoin and high-performance computing clients.  HIVE’s twin-turbo engine infrastructure-driven by hashrate services and GPU-accelerated AI computing-delivers scalable, environmentally responsible solutions for the digital economy.

For more information, visit hivedigitaltech.com, or connect with us on:

X: https://x.com/HIVEDigitalTech
YouTube: https://www.youtube.com/@HIVEDigitalTech
Instagram: https://www.instagram.com/hivedigitaltechnologies/
LinkedIn: https://linkedin.com/company/hiveblockchain

Advertisement

On Behalf of HIVE Digital Technologies Ltd.

“Frank Holmes”
Executive Chairman

For further information, please contact:

Nathan FastDirector of Marketing and Branding
Frank HolmesExecutive Chairman
Aydin KilicPresident & CEO

Advertisement

Tel: (604) 664-1078

¹ Claim of efficiency relates to research methods

Neither the TSX Venture Exchange nor its Regulation Services Provider (as that term is defined in policies of the TSX Venture Exchange) accepts responsibility for the adequacy or accuracy of this news release.

Forward-Looking Information

Advertisement

Except for the statements of historical fact, this news release contains “forward-looking information” within the meaning of applicable Canadian securities laws, which may include but is not limited to statements regarding: the performance of the BUZZ AI Cloud platform in Asunción, Paraguay; the ability to replicate and scale this performance; the benefits and advantages of power supply and Internet connectivity in Paraguay, the reorientation of the Swedish facilities to HPC standards;  the expected deployment, timing, capacity, and expansion of BUZZ HPC’s GPU-accelerated infrastructure in general; and any other future-oriented statements. Forward-looking information is based on current expectations, estimates, forecasts, and projections, as well as management’s beliefs and assumptions, including that the benefits of the operations in Paraguay can be replicated and scaled, infrastructure will be deployed on the expected timelines and within budget across all sites, demand for AI computing will continue to grow, and regulatory requirements will remain consistent with current expectations, and other related risks as more fully set out in the Company’s disclosure documents under the Company’s filings at www.sec.gov/EDGAR and www.sedarplus.ca.

Forward-looking information involves known and unknown risks, uncertainties, and other factors that may cause actual results, performance, or achievements to differ materially from those expressed or implied by such forward-looking information. Such factors include, but are not limited to the following risks: deployment timelines may change; costs may exceed expectations; performance expectations may not be achieved; demand for AI infrastructure may be lower than anticipated; partnerships or regulatory approvals may not materialize as expected; and the risk factors described in the Company’s continuous disclosure documents available on SEDAR+ at www.sedarplus.ca. Readers are cautioned not to place undue reliance on forward-looking information. The Company disclaims any obligation to update or revise any forward-looking information, whether as a result of new information, future events, or otherwise, except as required by law.

Risk & affiliate notice: Crypto assets are volatile and capital is at risk. This article may contain affiliate links. Read full disclosure

Advertisement

Source link

Continue Reading
Click to comment

You must be logged in to post a comment Login

Leave a Reply

Crypto World

Claude Managed launches in public beta

Published

on

Claude Managed launches in public beta

AI agents built on Anthropic’s Claude now have a hosted production infrastructure behind them, as the company launched Claude Managed Agents in public beta on April 8, handling the sandboxing, state management, credential handling, and error recovery that previously took engineering teams three to six months to build before writing a single line of agent logic.

Summary

  • Claude Managed Agents is available now on the Claude Platform at $0.08 per runtime hour plus standard Claude model usage costs; an agent running around the clock costs approximately $58 per month in runtime before token costs, and the service runs exclusively on Anthropic’s infrastructure
  • Early adopters already in production include Notion, which delegates coding, slides, and spreadsheet tasks to Claude in parallel across dozens of simultaneous sessions; Rakuten, which deployed specialist agents across product, sales, marketing, finance, and HR, each live in under a week; and Asana, whose CTO says the company shipped advanced features “dramatically faster” than prior methods allowed
  • Two features are in research preview: the ability for agents to create additional sub-agents for complex tasks, and an automatic prompt quality enhancement that improved structured file generation success rates by up to 10 points in internal testing

Anthropic’s @claudeai account announced the launch on April 8 at 5:14 PM ET, drawing 5.09 million views. The service is built around what Anthropic calls a brain versus hands design philosophy: Claude itself is the reasoning layer, while each session runs in a disposable, isolated Linux container that handles code execution, file manipulation, and tool calls. When the next Claude model ships, the infrastructure does not need to be rebuilt. The brain upgrades and the hands remain the same.

Pricing is usage based. The $0.08 per runtime hour applies to the session; standard Claude token pricing applies to model usage on top of that.

Advertisement

The deployment patterns across Notion, Asana, and Rakuten illustrate three distinct enterprise use cases. Notion integrated Claude directly into workspaces, allowing engineers to ship code and knowledge workers to generate presentations and websites without leaving the platform, running dozens of parallel tasks while teams collaborate on outputs simultaneously. Asana built what they call AI Teammates, agents embedded in project management workflows that pick up assigned tasks, draft deliverables, and hand back outputs for human review. Rakuten stood up agents across five business functions, each plugged into Slack and Teams, accepting task assignments and returning structured deliverables, with each function live in under a week. Sentry took a different path, pairing its existing debugging agent with a Claude-powered counterpart that writes patches and opens pull requests autonomously from a flagged bug to a completed pull request with no human intervention.

What Developers Need to Know Before Building

Developers define the agent by specifying the model, system prompt, tools, MCP server connections, and guardrails, then configure a cloud environment with pre-installed packages and network access rules. Anthropic’s infrastructure handles tool orchestration, context management, checkpointing, and crash recovery. Sessions persist through disconnections, a practical requirement for complex workflows. The one significant constraint is that the service runs only on Anthropic’s infrastructure and is not currently available through Amazon Bedrock or Google Vertex AI, which matters for organizations with multi-cloud strategies.

Why This Launch Matters for the Broader AI Market

As crypto.news has reported, the AI integration driving enterprise decisions in 2026 increasingly determines headcount, and the operational overhead that Claude Managed Agents eliminates has been a significant barrier to adoption for teams without specialist DevOps resources. As crypto.news has noted, the AI infrastructure buildout, of which Anthropic’s agent platform is a direct example, is one of the primary drivers of capital allocation decisions that have ripple effects across crypto-adjacent AI token markets. The multi-agent coordination feature, which allows agents to spawn sub-agents for complex tasks, is in research preview, with early access available through the Claude Platform console.

Advertisement

Source link

Continue Reading

Crypto World

TON’s Catchain 2.0 Delivers Sub-Second Finality, Shortening Latency

Published

on

Crypto Breaking News

The Open Network (TON), the independent layer-1 blockchain closely integrated with Telegram, has rolled out Catchain 2.0, dramatically shortening block times to 400 milliseconds. The upgrade is designed to push settlement speeds toward real-time for both payments and trades, while enabling decentralized applications to run with performance closer to traditional apps.

According to TON’s announcement, payment transactions now settle in about one second, and trades settle in near real time. The upgrade strengthens TON’s position as a platform aiming to blend messaging with on-chain functionality, a path already underscored by its ongoing Telegram integration. The update comes alongside an inflationary shift in TON’s token economics: annual inflation is projected to rise six-fold, to roughly 3.6% from about 0.6%, driven by the increased rate of block production.

“More blocks mean more validator rewards, which create stronger staking incentives and bring more TON into the network,” TON stated in its release. The Catchain 2.0 upgrade builds on TON’s Catchain consensus architecture, a BFT-style algorithm first proposed in 2020, and brings near-instant settlement to a network already embedded in an app ecosystem with approximately 1 billion users globally.

Market data captures a snapshot of how the upgrade is being received. TON was trading up about 2.3% to roughly $1.28 on Thursday, with volume around $130 million and a market capitalization near $3.17 billion, according to CoinMarketCap. Observers noted a surge in activity on TON’s network following the upgrade, including spikes in transactions per second tracked by TON Explorer, underscoring the immediate demand for faster settlement and more responsive smart-contract activity.

Advertisement

The announcement frames Catchain 2.0 as a natural evolution of TON’s thrust to merge everyday communications with on-chain finance, a vision that has been reinforced by Telegram’s growing crypto toolkit. In February, Telegram added self-custodial vaults to its in-app wallet, enabling users to earn yield on Bitcoin, USDT and ETH. Earlier this month, the wallet extended into perpetual futures trading, launching a new feature set in collaboration with the perpetual DEX Lighter. The integration enables TON-based payments and on-chain interactions directly within Telegram’s user interface, broadening the potential scale of user adoption and on-chain activity.

Key takeaways

  • Block time slashed to 400 milliseconds. Catchain 2.0 delivers substantially faster block finality, aiming to improve throughput and responsiveness for both financial transactions and developer applications.
  • Settlement accelerates to near real-time. Payments settle in about one second; trades settle in real time, enabling a smoother user experience for rapid on-chain exchanges.
  • Inflation expected to rise to 3.6%. The increase from roughly 0.6% reflects higher block production and the ongoing minting/burning dynamics within TON’s ecosystem.
  • Stronger staking incentives. More blocks translate into more validator rewards, reinforcing incentives to run validators and participate in securing the network.

Catchain 2.0: what changes and why it matters

At the core of the upgrade is TON’s Catchain consensus algorithm, a mechanism designed to achieve Byzantine fault tolerance while maintaining speed and finality. By accelerating block production, Catchain 2.0 effectively raises throughput across the network, which has several practical implications for users and developers. First, faster blocks reduce the latency between submitting a transaction and its confirmation, a critical factor for payment rails and decentralized finance (DeFi) applications that rely on quick settlement to minimize front-running and slippage. Second, the higher block rate inflates the expected rewards for validators, potentially strengthening the security of the network through deeper staking participation and a larger base of committed validators.

The inflationary shift, while potentially dilutive in the short term, is positioned by TON as a byproduct of increased activity and network security. The organization argues that the higher issuance supports a more robust staking economy, which can, in turn, bolster long-run network reliability and validator health as adoption grows. Investors and builders should weigh the inflationary impact against the benefits of faster settlement and a more responsive ecosystem, particularly as TON deepens its ties with Telegram’s user base and integrated financial features.

Telegram: turning messaging into a multi-asset financial channel

The upgrade arrives amid a broader narrative: TON’s alignment with Telegram is not merely cosmetic. The Telegram integration is positioned to enable users to send TON-enabled crypto payments within chats, bridging everyday communication and on-chain activity. The platform’s wallet features have evolved to offer in-app yield opportunities across major assets, including BTC, USDT, and ETH, and the ecosystem already supports perpetual futures trading through Lighter within the Telegram app. This progression points to a broader strategy of embedding crypto functionality into a widely used messaging interface, lowering the friction for mainstream users to engage with digital assets and on-chain commerce.

Pavel Durov, co-founder of Telegram, has highlighted how real-world restrictions and VPN workarounds in certain jurisdictions—such as Iran and Russia—have driven users to seek more resilient, open channels for communication. The TON-Telegram integration exemplifies a complementary model: users can exchange value alongside messages, with the possibility of automated payments and more sophisticated DeFi interactions embedded into the chat experience. For builders, this signals a shift toward app-layer ecosystems where identity, messaging, and asset transfer are increasingly interwoven.

Advertisement

Market response and next steps for TON

From a market perspective, TON’s price and on-chain activity suggest cautious enthusiasm for Catchain 2.0. The token’s modest near-term gain aligns with a broader pattern of traders evaluating how faster settlement and higher block production could influence user uptake, validator participation, and overall network throughput. The surge in on-chain activity reported by TON Explorer after the upgrade offers a tangible signal that developers and users are experimenting with new throughput capabilities and real-time interactions across the TON ecosystem.

Beyond immediate price moves, the key questions for investors and developers center on the durability of the new throughput gains, the sustainability of the higher inflation regime, and the extent to which Telegram’s in-app crypto features catalyze meaningful, recurring usage. Will higher staking rewards translate into deeper validator participation, and how will that impact network security and governance over time? How quickly will the on-chain experiences inside Telegram translate into real-world transaction volumes, merchant integrations, or consumer wallets?

Analysts will also be watching how Catchain 2.0 scales with continued ecosystem support. The near-term trajectory will depend on the balance between attracting new users through Telegram’s reach and maintaining robust validator participation to preserve the benefits of faster finality. In the meantime, developers can start leveraging the improved throughput to experiment with more sophisticated DeFi primitives, cross-chain liquidity, and real-time settlement use cases that were previously limited by latency.

What remains uncertain and what to watch next

While the upgrade delivers clear technical and user-facing benefits, several uncertainties deserve attention. The sustainability of the 3.6% inflation target hinges on adoption rates and the ongoing cycle of block production. The pace at which Telegram-integrated features translate into measurable user engagement and on-chain value remains to be seen, as does how regulatory developments may shape in-app crypto features and wallet services. Market participants will want to monitor validator health, network security metrics, and any changes in staking participation as Catchain 2.0 matures.

Advertisement

In sum, TON’s Catchain 2.0 represents a meaningful step toward faster, more interactive on-chain experiences embedded in a widely used messaging platform. For traders and developers, it signals a broader opportunity: a more responsive, scalable environment for payments, DeFi, and user-centric apps that live at the intersection of daily communication and digital assets. As TON continues to evolve its ecosystem—balancing security, inflation dynamics, and user adoption—the coming quarters will reveal how deeply this integration can redefine mainstream crypto usage.

Readers should watch for updates on validator participation, new application experiments on TON’s mainnet, and any material shifts in on-chain activity as Telegram-enabled features gain traction in real-world usage.

Risk & affiliate notice: Crypto assets are volatile and capital is at risk. This article may contain affiliate links. Read full disclosure

Advertisement

Source link

Continue Reading

Crypto World

Pyth Network Launches Data Marketplace For Price Feeds Across Asset Classes

Published

on

Data

Pyth Network, a blockchain data oracle provider, is launching a platform for financial institutions to publish and monetize their market data across blockchain networks. 

The Pyth Data Marketplace will initially support datasets for spot foreign currency exchange markets (FX), precious metals and crude oil swaps, while allowing publishers to retain “full control” over the data they share, according to Thursday’s announcement.

Seven new institutional data providers will publish price feeds on the marketplace at launch, the announcement said.

Data
Pyth’s price feeds span equities, commodities, precious metals and FX. Source: Pyth Network

These include stock exchange Euronext, data provider Exchange Data International, asset manager Fidelity Investments, financial exchange OTC Markets Group, Singapore Exchange FX and the Tradeweb trading platform. 

The announcement reflects how blockchain technology can democratize access to financial data, which has traditionally been controlled by a handful of service providers who charge exorbitant fees for high-quality market pricing data.

Advertisement

Related: Polymarket expands into equities and commodities with Pyth price feeds

Pyth to enable customers to “pull” data rather than traditional “push”

Pyth’s data pull model allows customers to pay for market data on demand, instead of traditional push-based oracle models that force users to pay for entire datasets, which they may or may not need.

This reduces the cost for the end user, according to Michael James, the head of institutional business development at Douro Labs, the main developer behind the Pyth Network.

Traditional service providers monopolize the $50 billion financial data industry, James told Cointelegraph at Consensus 2025. That is now being challenged by new emerging blockchain alternatives like Pyth and Chainlink.

Advertisement

“These data vendors have no competition in traditional finance, and so they have all the pricing power in the world,” he said.

Banks, hedge funds, trading firms and other financial institutions are forced to buy this financial data for “compliance” reasons, James added. 

Data
The market share of different blockchain oracle providers. Source: DeFiLlama

In August 2025, the US Department of Commerce selected Pyth and blockchain oracle provider Chainlink to publish economic data onchain.

Pyth was initially selected to publish quarterly gross domestic product (GDP) data, including five years of historical GDP figures, according to a previous announcement from the oracle provider.

However, Pyth anticipates adding support for more government economic data sets in the future. 

Advertisement

Magazine: Can blockchain solve its oracle problem?