NeuroSpark AI has added AMD Instinct GPUs on Oracle Cloud Infrastructure to run AI inference workloads, a move that underscores how fast-growing model-serving demand is widening the market beyond Nvidia and giving customers more options on cost and capacity.
NeuroSpark adds AMD Instinct GPUs on Oracle Cloud
The expansion matters because inference, not just training, is becoming the larger and more commercially important slice of AI spending as labs and enterprises push models into production. By pairing AMD’s Instinct accelerators with ROCm software on Oracle’s bare-metal cloud, NeuroSpark is betting it can serve advanced workloads with lower-cost hardware choices and more flexibility across model types.
Sentiment Indicators
In an AMD blog post dated Oct. 23, NeuroSpark said the additional hardware expands the compute capacity it can offer to AI labs, AI-native companies and enterprises using large language models, video generation, text-to-speech and embedding systems. The company said it manages the path from model to silicon across multiple hardware types and uses open-source ROCm software to tune workloads for customer-specific requirements.
The post also said NeuroSpark has ported several models to AMD infrastructure, including GLM-5.3 and frontier open-source systems such as Kimi K3, MiniMax M3 and Reflection’s Beam. Those claims were made in an AMD-published profile and were not independently verified in the material provided.
For AMD, the deal adds another example of its push to win share in AI inference, where buyers are increasingly sensitive to total cost of ownership, software support and deployment flexibility rather than raw chip performance alone. That is a key investor focus as AMD tries to narrow the gap with Nvidia, whose shares fell 8.1% to $204.64 on Oct. 30, while AMD slipped 12.7% to $504.80 after a sharp run earlier in the year.
Oracle also stands to benefit as its cloud infrastructure becomes a deployment layer for alternative AI accelerators, a potentially important revenue lever as hyperscalers and specialized AI infrastructure providers seek more supply choices. Oracle shares declined 6.3% to $130.67 on Oct. 30, with the stock still closely watched for its AI cloud bookings momentum and margins.
The broader backdrop remains favorable for chip suppliers and cloud providers able to secure scarce AI capacity. Data-center demand remains strong, while the AI sentiment gauge from Adalytica.com is in “Fear” at 18, reflecting elevated uncertainty even as companies continue to expand infrastructure.
The main question for investors is whether AMD can translate more design wins into durable inference share and whether Oracle can monetize that shift without margin pressure from hardware, energy and supply-chain costs. The next test will be whether more AI cloud operators follow NeuroSpark’s lead and whether customer adoption turns these deployment announcements into measurable revenue.
| Entity | Gains | Losses |
|---|---|---|
| AMD | ▲More AI inference share | ▼Nvidia’s dominance |
| Oracle Cloud Infrastructure | ▲More accelerator demand | ▼Cloud rivals with narrower hardware choice |
| NeuroSpark AI customers | ▲Lower-cost compute options | ▼Reliance on a single GPU stack |
| Nvidia | ▲Ongoing market lead, but pressure rises | ▼Some inference workloads shifting to AMD |
