AI InfrastructureSecurity

On-Prem Frontier AI for Banks and Government | Anand Kashyap, Fortanix | TFiR

0

Regulated enterprises — banks, healthcare systems, government agencies — want access to frontier-class AI models, but their compliance and security postures make sending sensitive workloads to external cloud endpoints untenable. The risks are concrete: data interception in transit, unverifiable retention policies, and the possibility that proprietary data influences model training for competitors. Without a deployment path that keeps data inside a controlled environment, adoption stalls entirely.

In this interview on TFiR, Anand Kashyap, CEO and Founder at Fortanix, walks through why regulated industries are demanding on-premises frontier model deployment and how confidential computing closes the technical gap between AI lab requirements and enterprise data sovereignty.

Guest: Anand Kashyap, CEO and Founder at Fortanix
Show: The Agentic Enterprise
LinkedIn: Anand Kashyap
Watch the discussion on YouTube

Here is what every security architect and AI platform engineer in a regulated industry needs to know.

Technical Deep Dive

Q: Why are banks and government agencies refusing to send data to frontier AI cloud endpoints?

Anand Kashyap, CEO and Founder of Fortanix, explains that regulated enterprises in banking, healthcare, and government share a common concern: they cannot verify what happens to their data once it leaves their perimeter. Specific fears include interception in transit, data retention by the AI provider, and the risk that proprietary information — such as credit risk modeling data — could inadvertently benefit a competitor. A recent public controversy over whether mathematicians’ work influenced an OpenAI solution to a math problem has heightened these concerns, even when AI labs provide informal assurances.

“We don’t want to send all our data to your cloud instance because we don’t know how it’s being transmitted, what path it takes, or where it could be intercepted.”

Anand Kashyap, CEO and Founder, Fortanix

Q: Why do regulated enterprises insist on frontier models rather than accepting open-weight alternatives?

Kashyap is direct: enterprises in regulated sectors are not satisfied with open-weight models as a substitute for frontier-class capability. The performance gap between openly available models and frontier models from leading AI labs is material for high-stakes applications like credit risk or clinical decision support. The demand is specifically for frontier model weights deployed inside a controlled, preferably on-premises environment — not a compromise on model quality in exchange for data control.

“Enterprises want to run the best models. They don’t want just open-weight models; they also want frontier-class models running in their own controlled environments, preferably on-premises.”

Anand Kashyap, CEO and Founder, Fortanix

Q: How does confidential computing protect both the AI lab’s model weights and the enterprise’s data at the same time?

The core technical challenge in on-premises frontier AI deployment is bilateral: the AI lab needs assurance that its model weights are not exposed or exfiltrated, while the enterprise needs assurance that its data remains isolated and unobservable. Kashyap describes confidential computing as the mechanism that satisfies both requirements simultaneously, using hardware-enforced isolation to protect the execution environment. Fortanix’s confidential AI solution, built on confidential containers, provides this dual protection without requiring either party to trust the other’s infrastructure claims.

“Confidential computing allows us to do that. And with our confidential AI solution, built using confidential containers, it is now much easier.”

Anand Kashyap, CEO and Founder, Fortanix

Resources & Documentation

  • Fortanix, confidential computing and confidential AI platform for regulated enterprises

***

👇 Click to Read Full Raw Transcript

Swapnil Bhartiya: A lot of your customers are in regulated industry like banks and governments. How does this change the conversation when they want to run Frontier models next to their own data?

Anand Kashyap: Yeah. So if you look at banks, healthcare industry, government, which are our primary target and primary industry which is most interested in computational computing, what they’re worried about is they don’t want their data to be exposed. They don’t want their data to leak and go somewhere where they lose control of what the data is. If a bank is building some kind of credit risk software, they’re worried that their data could be used and maybe distributed with some competitor banks as well. So increasingly, what we’re seeing is that the banks, healthcare, government agencies, they’re asking the Frontier labs and the leading AI labs, can we take your model weights and just run it on our own? We don’t want to send all our data out to your cloud instance because we don’t know how it’s going out, what path it takes, where it can be intercepted. We have seen recent news about the controversy, for example, with the Navy Strokes math problem, where there was a lot of controversy around whether the work that the mathematicians were doing did that influence the work that OpenAI did to solve the math problem. Right. So in absence of certainty from the Frontier labs saying, yeah, we will not look at your data, will not retain your data, will not train your models with their data. Enterprises want more control. They want to be able to run the best models. And they don’t want just open weight models. They do want the Frontier class models and they want to run it in their own controlled environment, preferably on prem. And as they’re trying to do that, there needs to be a way by which the model weights are protected so that the Frontier labs can feel that, yeah, my model weights are not getting exposed. At the same time, the data of the enterprise, that needs to be protected as well. So confidential computing allows us to do that. And with our confidential AI solution built with confidential containers, now, it makes it much more easier.

How to Switch Inference Providers With AI Gateway Routing | Jon Alexander, Akamai | TFiR

Previous article