How to manage token costs while keeping agentic goals on track
Sponsored by Dell AI Factory with NVIDIA
As enterprises evolve from chatbots to AI agents, they're discovering an unexpected challenge: rising token costs. Because AI usage is billed by tokens, and agents typically consume far more than chatbots, tokenomics is quickly becoming a real business issue.
This is why where you run AI matters as much as which AI you run, explains Bharat Patel, a solution architect at Dell Technologies Customer Solution Center. Here, he explains tokenomics and details how Dell Deskside Agentic AI, part of the Dell AI Factory with NVIDIA, runs workhorse models locally to manage pay-per-token costs.
Learn how the Dell AI Factory with NVIDIA can power your way to AI here: https://dell.com/YourWayToAI
Transcript:
Tokenomics is not an engineering conversation anymore. It's a P&L conversation.
My name's Bharat Patel. I'm a solution architect at the Customer Solution Center here at Dell Technologies, where I get to showcase Dell Technology's newest innovations, the business outcomes they drive, and the strategic value that they can create.
Tokenomics comes from token economics. So imagine tokens are just words that AI uses to communicate. What happens is a token, which is the basic unit that is used for AI to run workload; it's used to read and write text, so the more you use, the more cost it would become. It becomes an economic conversation, so hence the word tokenomics.
Tokenomics is not an engineering conversation anymore. It's a P&L conversation. It's a mindset shift because if you think about organizations, they've been using chatbots. Old-school chatbots are like vending machines. You press a button and out comes a snack. Agentic AI, on the other hand, is a little more complicated. They're like your home office agents. They can think, they plan, they use tools, they operate autonomously, but they run for extended periods of time. Chatbots only consume hundreds of tokens, whereas an agentic AI workflow can consume hundreds of thousands of tokens or even millions.
Now users are thinking, what can we do next? And that's why agentic AI has come about. From an agentic AI perspective, you are now having autonomous agents not go do simple tasks, but do more complicated tasks. Those complicated tasks consume more tokens. If they consume more tokens, then your bill is going to get higher.
This shift means that the entire supporting ecosystem, the hardware, the software, the security, now has to grow up with this new agentic AI world. This is no longer just an IT conversation. It's a whole company conversation. This requires new architecture for the work itself. You need trusted data, governance, infrastructure that's closer to make real-time decisions.
Where Dell and NVIDIA come into the picture is through Dell's Deskside Agentic AI solutions where you can build and run secure autonomous agents right at your desk. Dell's Deskside Agentic AI is powered by Dell's high-performance workstations and NVIDIA NemoClaw. Those workstations could be Dell Pro Precision Towers or the Dell Pro Max with the GB10 or even the Dell Pro Max with the GB300. This allows organizations to build and deploy agents and govern them with privacy controls. Think of Dell Deskside Agentic AI as a clear path from "let's try this," to "this is running in production." It converts variable cloud token costs into a controlled infrastructure investment.
The Dell AI factory with NVIDIA now extends to the edge and deskside. So you can start small, validate your use cases, and then grow into the infrastructure that we provide. It lets you run AI agents right at your site on hardware you own, not cloud servers you're renting. It's the same tools, the same playbook, but a bigger stage.
Here's what leaders who are exploring agentic AI need to keep in mind. Autonomy is the promise of agentic AI, but use cloud as a specialist when you need to, but run agentic AI on a local system as your everyday workhorse. Leaders should be more focused, not necessarily on the cost per token. They should be more focused on cost per outcome.
Where you run AI is just as important as which AI you run. The Dell AI factory with NVIDIA has the pieces for you to be successful in your organization from the edge, the deskside, the data center, all the way to the cloud. Leaders have to remember, start local, govern early, and scale smart.
Sponsored by Dell AI Factory with NVIDIA
As enterprises evolve from chatbots to AI agents, they're discovering an unexpected challenge: rising token costs. Because AI usage is billed by tokens, and agents typically consume far more than chatbots, tokenomics is quickly becoming a real business issue.
This is why where you run AI matters as much as which AI you run, explains Bharat Patel, a solution architect at Dell Technologies Customer Solution Center. Here, he explains tokenomics and details how Dell Deskside Agentic AI, part of the Dell AI Factory with NVIDIA, runs workhorse models locally to manage pay-per-token costs.
Learn how the Dell AI Factory with NVIDIA can power your way to AI here: https://dell.com/YourWayToAI
Transcript:
Tokenomics is not an engineering conversation anymore. It's a P&L conversation.
My name's Bharat Patel. I'm a solution architect at the Customer Solution Center here at Dell Technologies, where I get to showcase Dell Technology's newest innovations, the business outcomes they drive, and the strategic value that they can create.
Tokenomics comes from token economics. So imagine tokens are just words that AI uses to communicate. What happens is a token, which is the basic unit that is used for AI to run workload; it's used to read and write text, so the more you use, the more cost it would become. It becomes an economic conversation, so hence the word tokenomics.
Tokenomics is not an engineering conversation anymore. It's a P&L conversation. It's a mindset shift because if you think about organizations, they've been using chatbots. Old-school chatbots are like vending machines. You press a button and out comes a snack. Agentic AI, on the other hand, is a little more complicated. They're like your home office agents. They can think, they plan, they use tools, they operate autonomously, but they run for extended periods of time. Chatbots only consume hundreds of tokens, whereas an agentic AI workflow can consume hundreds of thousands of tokens or even millions.
Now users are thinking, what can we do next? And that's why agentic AI has come about. From an agentic AI perspective, you are now having autonomous agents not go do simple tasks, but do more complicated tasks. Those complicated tasks consume more tokens. If they consume more tokens, then your bill is going to get higher.
This shift means that the entire supporting ecosystem, the hardware, the software, the security, now has to grow up with this new agentic AI world. This is no longer just an IT conversation. It's a whole company conversation. This requires new architecture for the work itself. You need trusted data, governance, infrastructure that's closer to make real-time decisions.
Where Dell and NVIDIA come into the picture is through Dell's Deskside Agentic AI solutions where you can build and run secure autonomous agents right at your desk. Dell's Deskside Agentic AI is powered by Dell's high-performance workstations and NVIDIA NemoClaw. Those workstations could be Dell Pro Precision Towers or the Dell Pro Max with the GB10 or even the Dell Pro Max with the GB300. This allows organizations to build and deploy agents and govern them with privacy controls. Think of Dell Deskside Agentic AI as a clear path from "let's try this," to "this is running in production." It converts variable cloud token costs into a controlled infrastructure investment.
The Dell AI factory with NVIDIA now extends to the edge and deskside. So you can start small, validate your use cases, and then grow into the infrastructure that we provide. It lets you run AI agents right at your site on hardware you own, not cloud servers you're renting. It's the same tools, the same playbook, but a bigger stage.
Here's what leaders who are exploring agentic AI need to keep in mind. Autonomy is the promise of agentic AI, but use cloud as a specialist when you need to, but run agentic AI on a local system as your everyday workhorse. Leaders should be more focused, not necessarily on the cost per token. They should be more focused on cost per outcome.
Where you run AI is just as important as which AI you run. The Dell AI factory with NVIDIA has the pieces for you to be successful in your organization from the edge, the deskside, the data center, all the way to the cloud. Leaders have to remember, start local, govern early, and scale smart.