Hive's Q2 2026 Earnings Call: Contradictions Emerge on AI Cloud Strategy, Hashrate Outlook, Paraguay Expansion, and HPC Capacity

Monday, Nov 17, 2025 10:09 pm ET2min read
Aime RobotAime Summary

-

reported Q2 2026 revenue of $87., up 286% YoY, driven by mining and AI cloud expansion.

- The company plans to scale Buzz AI cloud to 11,000 GPUs and target $225M ARR by 2026, leveraging Paraguay's low-cost hydro power.

- Management emphasized disciplined ROIC (18% annualized) and vendor financing for 6,000 GPU deployment, avoiding major capital raises.

- Paraguay's government allocated 100MW additional power, supporting HIVE's dual-engine $750M ARR target combining mining and HPC by 2026.

Date of Call: November 14, 2025

Financials Results

  • Revenue: $87.3M for the quarter, up from $22.6M YOY and up from $45.6M sequentially
  • EPS: Basic loss of $0.07 per share, compared to ~$0.00 in the prior year comparable quarter
  • Gross Margin: 49%, compared to 2% in the prior year comparable quarter
  • Operating Margin: 49%, up from 35% sequentially (quarter-over-quarter improvement)

Guidance:

  • Target to reach ~35 EH and roughly $550M ARR from mining by end of 2026 if incremental 100 MW is deployed
  • Buzz (AI/HPC) target to scale to ~11,000 GPUs and $140M ARR for GPU cloud plus ~$85M ARR from hyperscaler colo, targeting ~$225M for Buzz by end-2026
  • Combined dual-engine target of ~$750M ARR by end-2026 (mined revenue + HPC/colo)
  • 6,000 Blackwell GPUs to be brought online in tranches (vendor lease-to-own financing; first clusters in Q1–Q4 2026)
  • Incremental 100 MW in Paraguay approved; incremental build cost ~$250k/MW

Business Commentary:

  • Revenue Growth and AI Expansion:
  • HIVE Digital Technologies reported revenue of $87.3 million for the second quarter of fiscal 2026, up from $22.6 million in the same quarter last year.
  • The growth was driven by strong performance in Bitcoin mining and significant expansion in AI cloud services.

  • Operational Capacity and Efficiency:

  • The company's operational capacity in Paraguay was increased by 10 exahash to 25 exahash, with over 440 megawatts of green energy.
  • This increase in capacity was enabled by the country's surplus electricity and efficient ASIC mining operations.

  • HPC and AI Cloud Growth Strategy:

  • HIVE's AI cloud segment is poised for growth, with plans to double its GPU cluster from 5,000 to 11,000 by the end of 2026.
  • This expansion is supported by partnerships with companies like Bell Canada and Dell, facilitating sovereign data strategies.

  • Bitcoin Mining and Hash Rate Trends:

  • HIVE's Bitcoin mining operations achieved 10 Bitcoin per day, despite a difficulty increase to 152 trillion.
  • The company's ability to maintain profitability in Bitcoin mining is attributed to efficient ASIC management and low-cost green energy in Paraguay.

    Sentiment Analysis:

    Overall Tone: Positive

    • Management called the quarter "record-breaking" with $87.3M revenue, 25 EH installed, ~400M ARR mining run rate, 18% annualized ROIC and announced targets of ~11,000 GPUs for Buzz and combined ~$750M ARR by end-2026, emphasizing disciplined ROIC and vendor financing to fund growth.

Q&A:

  • Question from Darren Aftahi (Roth): What is the assumed CapEx per megawatt to retrofit/greenfield to tier three, and on the AI cloud financing do you plan to buy GPUs at lease end and what useful life are you assuming for GPUs?
    Response: Conversion CapEx: Boden ~$25M and Toronto ~$40M; Bell colo requires no significant CapEx; GPUs are lease-to-own with a de minimis buyout and management conservatively assumes a ~5-year useful economic life (they cited ~50% conservative residual after 3 years).

  • Question from Mike Grondal (Northland): How many of the 6,000 GPUs are contracted today, what visibility do you have, average term, and demand profile?
    Response: GPU rollout is staged (500 ordered arriving before Christmas; cadence: ~1,000 in Q1 2026, 1,000 in Q3, 4,000 in Q4); existing monetization mixes are 3–12 month terms while the base-case modeling uses two-year long-term contracts for conservative ARR visibility.

  • Question from Bill Papanastasio (Stifel): Views on where hash rate and economics could trend given AI HPC expansions and capital strategy between HPC and mining?
    Response: Primary principle: will only scale ASICs if they meet ROIC targets (~1–1.5 year ROI); mining break-even is ~ $22 hash-price and they expect difficulty/hash-price fluctuations but will prioritize ROIC and use vendor financing for GPU growth.

  • Question from Chris Brendler (Rosenblatt): With phase three now complete, should gross margins continue to improve as more efficient hash rate and Paraguay power ramp?
    Response: Yes — additional efficient machines (15 J/TH) and low-cost Paraguay hydro reduce average hash cost and should further improve gross mining margins as new capacity stabilizes.

  • Question from Fedor Shavelin (B. Riley): Timeline and phasing for the New Brunswick data center conversion, and if the colocation agreement doesn't proceed can capacity be repurposed and how long to finance GPUs?
    Response: New Brunswick conversion is planned for hyperscaler colo with engineering/land work targeted to enable construction starting in Q1 (timing updates to follow); if colo terms change the site can be repurposed for AI workloads, and GPU procurement is largely mitigated by vendor lease-to-own financing so significant external capital raises are not required for the 6,000-GPU plan.

  • Question from Joe Vaffe (Canaccord): How much additional power is available from Itaipu/Paraguay and what is the government's stance on data centers/AI?
    Response: Paraguay (Itaipu region) offers very large low-cost hydro capacity (~14–15 GW total) and the government is supportive — HIVE was recently awarded an additional 100 MW allocation, and management emphasized ongoing close collaboration with Paraguayan leaders to expand data center activity.

Contradiction Point 1

AI Cloud Strategy and GPU Acquisition

It involves the company's strategy regarding the acquisition and useful life of GPUs for their AI cloud strategy, which is crucial for their growth and competitive positioning in the AI market.

What is the assumed CapEx per megawatt for retrofitting or building new Tier 3 data centers? How do you plan to handle GPU acquisition and useful life in your AI cloud strategy? - Darren Aftahi (Roth)

2026Q2: The acquisition of GPUs will be handled through lease-to-own agreements, with ownership transfer at a minimal cost at the end of the term. The useful life of GPUs is conservatively estimated at five years, but they could last longer based on historical data and ongoing demand. - Aydin Kilic(CEO)

Has demand for the AI cloud business changed since the DeepSeek announcement? - Darren Aftahi (ROTH Capital)

2025Q3: We'll probably talk about that in the next call, and the AI -- the AI cloud division is focused on next-generation video compute, GPU-based compute for the next generation of video compute, consolidating, and to work with hyperscalers to deliver the best service we can. - Frank Holmes(Executive Chairman)

Contradiction Point 2

Hashrate and Profitability Outlook

It pertains to the company's expectations regarding hashrate and profitability, which are crucial for investors and stakeholders.

Can you discuss the potential trends in hash rate and economics as the industry shifts to AI HPC, and your strategy moving forward? - Bill Papanastasio (Stifel)

2026Q2: The current difficulty is at 152 trillion, with a downward adjustment expected in the next epoch. The ratio of Bitcoin price to hash price is expected to stabilize, with a downward trend anticipated. - Aydin Kilic(CEO)

What were Hive's Q1 2026 financial results? - Eitan Killick (Hive Digital Technologies)

2026Q1: Adjusted EBITDA increased to $44.6 million in Q1 2026 compared to $(4.3) million in Q1 2025. The increase was primarily driven by an increase in revenue from Bitcoin mining. - Darcy DuBarris(CFO)

Contradiction Point 3

GPU Acquisition Strategy and Timing

It involves differences in the strategy and timing for GPU acquisition, which is crucial for HIVE's AI cloud strategy and revenue projections.

What is the capital expenditure per megawatt for upgrading or building Tier 3 data centers? How do you plan to manage GPU acquisition and their useful life within the AI cloud strategy? - Darren Aftahi(Roth)

2026Q2: The acquisition of GPUs will be handled through lease-to-own agreements, with ownership transfer at a minimal cost at the end of the term. The useful life of GPUs is conservatively estimated at five years, but they could last longer based on historical data and ongoing demand. - Aydin Kilic(CEO)

What is the strategy for the Toronto HPC expansion project? What is the financing plan? - Fedor (B. Riley)

2025Q4: The data center can be financed through various mechanisms, such as vendor financing and leasing. The budget acquisition cost is competitive, with a 9-month retrofit expected to triple current HPC capacity. - Aydin Kilic(CEO)

Contradiction Point 4

Paraguay Expansion Strategy and Opportunities

It highlights differing perspectives on the company's expansion strategy in Paraguay and its potential for future growth opportunities.

What is your strategy in Paraguay regarding the Itaipu Dam's power output and how Paraguay plans to use AI and data centers? - Joe Vaffe (Canaccord)

2026Q2: Itaipu is a significant source of green energy, with potential for Paraguay to become a data center hub. The government is proactively positioning the country for growth in this sector. The company is working closely with the government to support and grow data center infrastructure. - Aydin Kilic(CEO)

What are the details of the Paraguay expansion strategy, particularly opportunities for further regional growth? - Bill Papanastasiou (Stifel)

2025Q3: The team is focused on the 300 megawatt transformational project in Paraguay. Conversations are ongoing with existing miners in the region. While opportunities continue to be explored, the company remains focused on executing the current expansion, prioritizing it to ensure successful implementation. - Darcy Daubaras(CFO)

Contradiction Point 5

HPC Capacity and Demand

It involves differing statements about the capacity and demand for HPC resources, which impacts business strategy and revenue projections.

How many of the 6,000 GPUs are currently contracted? What is the visibility on these contracts? What is the average contract term? - Mike Grondal(Northland)

2026Q2: The first cluster of 1,000 GPUs is contracted and expected to come online by Q1 2026. - Aydin Kilic(CEO)

How is the demand for the 5.5 MW IT load evolving? - Mike (Northland)

2025Q4: There is strong demand with overcommitted GPUs. - Aydin Kilic(CEO)

Comments



Add a public comment...
No comments

No comments yet