-
How much GPU do you need to run an open model?
A practical map from GPU memory tiers to the Pendra models that actually fit, from a used 24GB card to unified-memory boxes and datacentre silicon.
-
Chat, images, embeddings and audio on Pendra
Pendra is more than a chat API. The same sovereign UK platform also generates images, creates text embeddings and transcribes audio, all behind OpenAI-compatible endpoints.
-
When your AI vendor can be switched off by a government
When the US government ordered Anthropic to suspend two of its most capable AI models worldwide, it showed that access to a foreign-run frontier model can be revoked overnight. For UK businesses that is a continuity risk, not just a privacy one.
-
Pendra joins NVIDIA Inception
Pendra has joined NVIDIA Inception, giving our sovereign UK AI inference platform closer access to NVIDIA technology, preferred GPU pricing and technical support as we grow.
-
Your data can sit in Europe and still reach Washington
Microsoft handed the unredacted names of Dutch regulators to the US House of Representatives under the CLOUD Act, even though the data sat on European servers. For UK organisations it is a reminder that data residency is not the same as data sovereignty.
-
Most UK companies can't see where their AI sends their data
A Financial Times survey of senior tech leaders at large UK companies found 61% do not fully understand how their data is handled when AI processes it overseas, and 75% see UK data leave the country through AI every week.