Verda is the new name for DataCrunch, the Helsinki-based GPU cloud that runs on renewable energy in Finland and Iceland and has quietly built one of the broadest Blackwell rate cards in Europe: B200, B300, GB300, alongside H200, H100, A100, RTX Pro 6000 and L40S. Its distinguishing feature is a published spot tier at 50% of on-demand across the board, plus a serverless container product.
packet.ai reaches similar prices with a different mechanism: Dynamic instances that share a card under scheduler-enforced isolation and are never pre-empted. This comparison lines up Verda's on-demand and spot columns against packet.ai's Dynamic and Dedicated.
Platform overview: packet.ai vs Verda
What Verda offers
Verda's published on-demand rates as of late August 2026: GB300 $8.62/hr, B300 $7.50/hr, B200 $6.11/hr, H200 SXM $4.00/hr, H100 SXM $3.25/hr, RTX Pro 6000 (96GB) $1.89/hr, A100 SXM 80GB $1.79/hr, A100 40GB $1.29/hr, L40S $1.37/hr, RTX 6000 Ada $1.04/hr, RTX A6000 $0.61/hr. Spot is 50% of on-demand on every card (B200 $3.06, H200 $2.00, RTX Pro 6000 about $0.95). Reserved terms from 1 to 24 months discount 2–25%.
- Per-second billing, no minimum commitment; instances from 1× to 8× on most cards
- Renewable-powered data centres in Finland and Iceland; ISO-certified; strong EU residency story
- Serverless containers priced separately (H100 $3.58/hr, L40S $1.51/hr) with spot at about half
- Confidential-computing variants of B200 and RTX Pro 6000 at a small premium
What packet.ai delivers
Verda's published spot tier, a flat 50% off every card, is the most aggressive discount structure in this whole series. packet.ai reaches similar numbers by a different route: Dynamic instances share a card under scheduler isolation, and unlike spot, they are never reclaimed. On the RTX Pro 6000 that mechanism actually wins outright: packet.ai Dynamic at $0.66/hr is 30% below Verda's spot price of about $0.95/hr, with no interruption risk at all.
On-demand to on-demand, packet.ai is cheaper on every shared card: RTX Pro 6000 65% less, L40S 33% less, A100 80GB 20% less, and B200 Dynamic 39% less than Verda's on-demand rate.
- Dynamic: scheduler-enforced multi-tenant, never pre-empted. RTX 6000 Pro at $0.66/hr, below Verda's own spot rate for the same card. B200 at $3.75/hr, 39% under Verda's on-demand.
- Dedicated: a whole card, single-tenant, 99% SLA, for workloads that need guaranteed exclusivity rather than shared scheduling. RTX 4090 at $0.39/hr, L40S at $0.92/hr, A100 80GB at $1.43/hr, B200 at $6.99/hr.
What Verda has that packet.ai does not: H100, H200 and B300 on the rate card today, a true spot tier for checkpointed training that can beat Dynamic on the largest cards, confidential-computing variants, and Nordic renewable-powered residency. If your workload is Hopper-class and can tolerate interruption, Verda's spot tier is genuinely hard to beat. If you need non-interruptible capacity at a price below anyone's spot rate, packet.ai's Dynamic tier does that on the RTX Pro 6000 specifically.
Billing is hourly with monthly commits up to 20% off. There are no platform fees and no ingress charges; egress is $0.04/GB. Up to 2 TB of local NVMe per node is included. Dynamic instances are SSH-ready in under five minutes, Dedicated in five to ten. Capacity is live in the US (California, Virginia, Texas, Oregon) and Europe (Frankfurt, Amsterdam, Paris, London, Dublin). H100 SXM, H200 and RTX 5090 are on the notify list.
Against Verda, packet.ai's on-demand Dynamic rates sit below Verda's spot on the RTX Pro 6000 ($0.66 versus about $0.95) and above it only on the B200 ($3.75 versus $3.06), where Verda's spot can be reclaimed and packet.ai's Dynamic cannot. On Dedicated versus on-demand, packet.ai is 20% cheaper on the A100 80GB and essentially tied on the B200.
packet.ai vs Verda at a glance
Published starting rates in USD per GPU-hour, on-demand unless noted. Percentages compare the first price in each cell.
| Category | packet.aiYOU | Verda |
|---|---|---|
| RTX 6000 Pro (96GB) | $0.66/hr Dynamic−65% | $1.89/hr on-demand · ~$0.95/hr spot |
| B200 (180–192GB) | $3.75/hr Dynamic · $6.99/hr Dedicated−39% | $6.11/hr on-demand · $3.06/hr spot |
| A100 80GB | $1.43/hr Dedicated−20% | $1.79/hr on-demand · ~$0.90/hr spot |
| L40S (48GB) | $0.92/hr Dedicated−33% | $1.37/hr on-demand · ~$0.69/hr spot |
| RTX 4090 (24GB) | $0.39/hr Dedicated | Not offered (RTX A6000 $0.61/hr) |
| H100 / H200 / B300 | H100 and H200 launching soon (notify list) | H100 $3.25/hr · H200 $4.00/hr · B300 $7.50/hr |
| Interruptible tier | None; Dynamic is shared, not pre-empted | Spot at 50% of on-demand, can be reclaimed |
| Billing | Hourly; monthly commits up to 20% off | Per second; reserved 1–24 months, 2–25% off |
| Serverless | Token Factory API (waitlist) | Serverless containers, separate rate card |
| Regions | US (4 sites) + EU (5 sites) | Finland, Iceland |
| Data transfer | No ingress fee; egress $0.04/GB | Not itemised in sources reviewed |
Detailed comparison
On-demand versus on-demand
Card for card at list price, packet.ai is lower everywhere the two overlap: RTX Pro 6000 $0.66 against $1.89 (65% less), L40S $0.92 against $1.37 (33% less), A100 80GB $1.43 against $1.79 (20% less). On the B200, packet.ai Dynamic at $3.75 is 39% under Verda's $6.11, while packet.ai Dedicated at $6.99 is 14% above it.
Verda has Hopper and B300 on the card today; packet.ai has notify lists. If your workload is H100-shaped, Verda's $3.25/hr H100 SXM is one of the better published prices in Europe.
Spot versus Dynamic
Verda's spot tier is the strongest published discount in this series: half price on every card, no exceptions. Against it, packet.ai Dynamic is cheaper on the RTX Pro 6000 ($0.66 versus about $0.95) and more expensive on the B200 ($3.75 versus $3.06), the A100 ($1.43 Dedicated versus about $0.90) and the L40S ($0.92 Dedicated versus about $0.69).
The difference is what you are buying. Spot can be reclaimed when Verda needs the capacity; Dynamic cannot, and Dedicated carries a 99% SLA. For checkpointed training, Verda spot is excellent value. For anything serving traffic, the comparison is Dynamic against on-demand, and packet.ai wins that.
Residency and sustainability
Verda's Finnish and Icelandic data centres run on renewable power and it is ISO-certified, which makes it a natural fit for European buyers with sustainability or residency requirements. packet.ai has five EU sites (Frankfurt, Amsterdam, Paris, London, Dublin) plus four in the US; a team serving both continents from one vendor has that option on packet.ai and not on Verda.
Serverless and confidential computing
Verda offers serverless GPU containers with their own rate card and confidential-computing variants of the B200 and RTX Pro 6000 for regulated workloads. packet.ai's serverless answer is the Token Factory API, on a waitlist, and it does not currently list confidential-computing SKUs.
Who should choose which
Choose packet.ai if
- You need a B200 on the self-serve rate card today, without a sales conversation
- You want Dynamic pricing — spot-tier economics without pre-emption risk
- Your workload is inference or any pipeline that cannot tolerate interruption
- You need European capacity across multiple EU countries rather than a single Finnish data centre
Choose Verda if
- You need an H100 SXM at a sub-hyperscaler price today, on-demand or spot
- You are running batch training workloads that checkpoint and can tolerate spot pre-emption
- Finnish or Nordic data-residency is a procurement requirement
- The 50%-off spot discount on the A100 PCIe at $0.73/hr fits your workload and budget
Conclusion
Verda is one of the best-priced European GPU clouds, and its 50% spot tier is genuinely aggressive. packet.ai is cheaper on every card at on-demand, is under Verda's spot on the RTX Pro 6000, and never pre-empts you. Use Verda spot for training you can restart; use packet.ai for everything that has to stay up.
Same silicon. Smarter economics.
Every rate on this page is a published starting price you can deploy against today. Dynamic launches in under five minutes; no credit card to start.
Additional resources
- packet.ai GPU pricing: the full Dynamic, Dedicated and Clusters rate card.
- packet.ai documentation: API, CLI and platform guides.
- How intelligent scheduling works: why Dynamic keeps peak performance at a lower price.
- Clusters and wholesale quotes: multi-node InfiniBand capacity at ~30% below retail.
- Rent GPUs on packet.ai: specs, pricing and workload fit for every card.