SambaNova's SN40L is a dataflow chip with a three-tier memory hierarchy, which lets one system hold several large models and swap between them quickly. Intel explored buying the company for about $1.6 billion in late 2025; the talks stalled, and in February 2026 SambaNova raised $350 million in a Vista-led round and signed an Intel partnership instead.
Key Takeaways
- What it is — Inference specialist. Dataflow chip that maps the model graph onto reconfigurable compute, with a three-tier memory of SRAM, HBM and DDR.
- Best for — Enterprises and governments that want fast inference on their own premises with many models per rack.
- Watch out — Intel's acquisition talks stalled; the company now depends on its February 2026 round and the Intel partnership.
At a glance
| Company | SambaNova Systems |
|---|---|
| Current chips | SN40L reconfigurable dataflow unit (RDU) |
| Architecture | Dataflow chip that maps the model graph onto reconfigurable compute, with a three-tier memory of SRAM, HBM and DDR |
| Memory | Three tiers per chip (on-chip SRAM, HBM and DDR), so one rack can hold many large models and switch between them |
| Scale-up | Rack-scale systems, sold as cloud capacity or installed on premises |
| Software | SambaNova Cloud API for open-weight models and an enterprise stack for private deployments |
| Where to get it | SambaNova Cloud, plus dedicated and on-premises deployments |
| Pricing | Per token on SambaNova Cloud; enterprise and on-premises deals quoted |
What changed in 2026
- Dec 2025. Reports say Intel is close to buying SambaNova for about $1.6 billion including debt; the talks later stall.
- Feb 2026. SambaNova raises $350 million led by Vista Equity Partners and signs a partnership with Intel.
Scores
Scored 1 to 5 on the same five axes as every vendor in the AI chip companies analysis.
| Performance | Scale-up | Software | Access | 2026 momentum | Total / 25 |
|---|---|---|---|---|---|
| 4 | 3 | 3 | 3 | 2 | 15 |
Verdict
A credible on-premises inference option when data residency rules out public APIs. Diligence the balance sheet as carefully as the benchmarks.
Alternatives
In the same lane (inference specialist):
- Cerebras: the fastest tokens per second on open-weight models, for agents and interactive products where latency is the product.
- Groq: low-latency, low-cost inference on popular open-weight models through a simple API.
In other lanes:
- Google TPU (hyperscaler custom silicon)
- AWS Trainium (hyperscaler custom silicon)
- AMD Instinct (merchant gpu)
- Tenstorrent (open hardware)
- Microsoft Maia (hyperscaler custom silicon)
Head-to-head comparisons
- Google TPU vs AWS Trainium vs Nvidia GPU: The three ways to train and serve frontier-scale models in 2026, scored on scale-up domain, software, access and price.
- Cerebras vs Groq vs SambaNova: The SRAM-first inference specialists compared on speed, token price, deployment options and company risk.
New to custom AI silicon? Our partner site CTAIO explains how TPUs, pods and GPU clusters differ in TPU vs GPU.
Frequently asked questions
What is SambaNova?
SambaNova's SN40L is a dataflow chip with a three-tier memory hierarchy, which lets one system hold several large models and swap between them quickly. Intel explored buying the company for about $1.6 billion in late 2025; the talks stalled, and in February 2026 SambaNova raised $350 million in a Vista-led round and signed an Intel partnership instead.
How do you get access to SambaNova?
SambaNova Cloud, plus dedicated and on-premises deployments. Pricing: Per token on SambaNova Cloud; enterprise and on-premises deals quoted.
What software does SambaNova run?
SambaNova Cloud API for open-weight models and an enterprise stack for private deployments.
Is SambaNova a good alternative to Nvidia?
A credible on-premises inference option when data residency rules out public APIs. Diligence the balance sheet as carefully as the benchmarks. It fits best for: enterprises and governments that want fast inference on their own premises with many models per rack. The main caveat: intel's acquisition talks stalled; the company now depends on its February 2026 round and the Intel partnership.
Sources
- SiliconANGLE: Intel to acquire SambaNova report
- DCD: SambaNova seeks funding after Intel talks stall
- The Star (Reuters): SambaNova raises $350 million
Facts verified 2026-10-31. Vendor performance figures are as stated by the vendor.
Ready to Find the Right AI Tools?
Browse our data-driven rankings to find the best AI tools for your team.