Meta won’t judge employees by how much they use AI when it comes to annual performance reviews, despite early efforts to drive AI adoption focusing on so-called token maxing.

The company has told employees that it “will not use AI adoption dashboards or token counts to evaluate impact,” according to  a report by The Information.

The announcement came in an internal memo from executives Maher Saba and Santosh Janardhan, which said that instead of measuring AI usage, “managers should look at output quality, velocity, problem complexity and scope taken on.”

This marks a culture change for Meta, where engineers had previously competed to consume the most AI tokens, displaying their scores on a leaderboard. Meta then discovered that its employees were being diverted from regular work because they were using AI to carry out additional tasks to boost their scores.

Amazon had similar results when it implemented a leaderboard to track AI use; it also found that some employees were trying to game the system by using AI to complete unnecessary tasks, and it has now deleted it.

The company, however, does also monitor employees’ use of AI for training purposes, in a program introduced in April, but this is information was not used to measure employee performance.

Meta had already started to look askance at the concept of using AI metrics as a tool to assess employees. Earlier this year, Chief Technology Officer Andrew Bosworth told employees in a memo that “nobody should be using AI tools just for the sake of using them,” adding that “token usage alone is not a measure of impact of any kind.”

Meta rolled out a new 30-billion-parameter AI model optimized to run on a PC or Mac with a single GPU on Monday, offering a way to run always-on agentic workflows locally rather than relying on the cloud.

The company has dubbed it Muse Glimmer. However, its hardware demands, including a GPU with a minimum of 24GB of VRAM, could make it difficult to justify for deployment at scale.

Although analysts and consultants agree that there is a tremendous enterprise appetite for running models locally, determining whether switching more systems from cloud to local makes fiscal sense is much more complex. 

The hardware costs are tricky to calculate even today, with the VRAM needed depending on the particular applications to be run. But the far bigger consideration is that there is no way to determine what RAM costs will look like over the next 12-18 months, and there is an identical lack of visibility into how cloud prices might increase during the same timeframe. 

That makes determining the better financial choice impossible.

Agents become capex, not opex

Noah Kenney, principal consultant at Digital 520, noted that since RAM costs have increased “exponentially” over the last 12 months, cloud AI providers are also going to have to increase their prices. Beyond that, IT needs to anticipate logistical issues; key questions to ask are, “How quickly can you scale? Can you even get the hardware?”

“Meta just made agents a capital expense instead of an operating one,” Kenney said. “For two years, enterprises have been trained to rent intelligence by the token from someone else’s data center. Muse Glimmer runs the agent on a GPU you own, on the desk, with the meter switched off. That is a direct shot at the business model that cloud AI vendors are built on, and it comes from the one player with no cloud API revenue to protect.”

In its post announcing the new model, Meta pointed out that it has aggressively slimmed it down to try to make it efficient and cost-effective. 

“At full precision, a 30-billion parameter model would require over 55 GB of memory — far more than any consumer GPU offers,” Meta said. “We use quantization techniques to compress the model’s weights to approximately 4-bit precision, shrinking the language model to under 20 GB. This leaves enough headroom for the model’s working memory, its KV cache, the perception encoder for image understanding, and the speculative decoding drafter to run simultaneously within a 24 GB or 32 GB envelope. We validated that this compression introduces minimal to no degradation on agentic tasks.”

More options for enterprises

Mike Wilkes, enterprise CISO at Aikido Security, said that Meta’s move is significant in that it starts to give enterprises more options. 

“The most important thing about Muse Glimmer is not that Meta has produced another capable model, it is that the economics and architecture of AI are beginning to move back toward the edge,” he said. “The financial comparison therefore becomes capital expenditure that can be amortized over several years versus an effectively perpetual per-token or per-request cloud operating expense.”

That means, he said, that an enterprise may rationally pay somewhat more for hardware if doing so gives it predictable AI costs, offline availability, control over model versions, freedom from sudden API pricing or access changes.

Independent cybersecurity and risk advisor Steven Eric Fisher also noted that the specs published by Meta don’t tell the full story.

The problem is that running locally versus in the cloud can generate a lengthy list of related expenses. 

“Agentic workloads [in the cloud] can amplify consumption through reasoning, retries, tool calls, context growth, and evaluation, while local deployment [also] introduces hardware, power, lifecycle, support, and utilization costs,” he said, adding that even the RAM requirements need a lot of context.

“Meta’s stated 24GB and 32GB memory targets demonstrate that Glimmer can be loaded and executed on comparatively accessible hardware, but that is not the same as having sufficient capacity for meaningful agentic workloads,” Fisher said, pointing out that once other factors are considered, practical memory requirements can move beyond the 32 GB available on an Nvidia RTX 5090 GPU.

“In enterprise terms, this still places Glimmer primarily in high-end developer, data science, or dedicated AI workstations rather than the standard corporate desktop or laptop,” he said.

Better ROI not guaranteed

Justin Greis, CEO of consulting firm Acceligence, agreed. 

“I wouldn’t assume that moving inference from the cloud to the endpoint automatically produces a lower total cost of ownership,” he said, noting that variable cloud costs would be traded for the price of  deployment, endpoint management, support, security, model updates, and potentially accelerated hardware refresh cycles for the local devices.

“Muse Glimmer is an important milestone because it makes local agentic AI technically viable. But technical viability and enterprise ROI are two different milestones,” Greis said. “I think Meta has crossed the first one. I do not think they have fully crossed the second one yet.”

However, Kenney argued that there are also other elements of the Meta rollout that make meaningful comparisons difficult. 

“It is worth remembering that the local model is quantized, compressed to roughly 4-bit precision, while the cloud APIs you are comparing against typically serve full-precision models, so this is not a pure apples-to-apples cost comparison,” he said. “The ROI question is not local versus cloud on price alone. It is also a question of whether a company is willing to use a quantized model for the specific task. A cheaper agent that needs more retries or human correction can erase its savings fast.”

In addition, a local model often neglects to include every service that a cloud provider typically delivers. 

“The cloud vendor was quietly handling updates, scaling, reliability, and security patching across your whole footprint. Bring the model in-house and every one of those becomes your problem, multiplied by every machine running it,” Kenney noted. “Most enterprises that consume AI as a service have no muscle for operating a fleet of local models, and that cost rarely gets adequate consideration in ROI conversations. The GPU is cheap. Patching a thousand of them is not.”

Sanchit Vir Gogia, chief analyst at Greyhound Research, also stressed that it can be difficult for IT to comprehensively anticipate the different cost variables. 

“Finance leaders are right to feel skeptical. An engine bought for one employee burns capital whether or not it runs. Meta has shown that a thirty-billion-parameter agent can run on a single machine, and that is a real engineering result. What it has not shown is that such an agent works reliably at enterprise scale, or that a fleet of them can be operated safely. Model fit and production fit are different claims. A laptop must still run the employee’s actual job,” Gogia said. “The economics turn on the incremental hardware premium, refresh timing and actual utilization, measured against the price of the remote inference being displaced.”

On the other hand, Arun Chandrasekaran, a distinguished VP analyst with Gartner, said that he found it “very interesting that they have decided to release a smaller model that operates on the edge” and especially liked Meta’s use of the popular Apache license. But he would have preferred that they had released more than just a model.

“Enterprise customers are asking for a car and Meta is delivering an engine,” Chandrasekaran said. “They should have built something more like a platform solution.”

Enterprises are moving aggressively into generative AI. On the surface, that seems like the right call. The technology is powerful, accessible, and increasingly embedded in how businesses build applications, automate processes, and support decision-making. A development team can connect an application to a large language model in days. A product team can add AI features in weeks. Business leaders see quick wins, faster innovation, and a path to modernizing nearly every part of the company.

These are the upsides everyone is talking about. The part we don’t discuss enough is the economic trap forming underneath all this convenience.

Most enterprises think of tokens as a technical billing detail. They are not. Tokens are the unit of economic dependency in generative AI. Every prompt, response, summarization, retrieval step, workflow action, and agent decision is measured and monetized through tokens. Tokens are not just part of the plumbing. They are the tollbooth between your enterprise and a provider’s intelligence platform. The more AI becomes central to your operations, the more power that tollbooth holds over your future costs.

Tokens are not just a pricing unit

A token is usually described as a chunk of text processed by a model. That is accurate enough for developers, but it misses the bigger issue for CIOs, architects, and corporate boards. In the enterprise, tokens are the mechanism by which AI capabilities are rented. They are the meter attached to the intelligence itself.

That distinction matters because token usage grows faster than most companies anticipate. A simple user prompt rarely remains simple in production systems. It can trigger retrieval from internal knowledge stores, multiple model calls, tool use, post-processing, policy checks, and agent loops. What appears to be a single transaction to the user may involve several layers of token consumption behind the scenes. As a result, enterprises often underestimate the true operating cost of AI-enabled systems, especially as those systems mature and spread across departments.

Today, those costs still feel manageable. In many cases, they feel surprisingly low. That is exactly why the trap is so dangerous.

The market is in a subsidy phase

Current token pricing is giving enterprises a false sense of comfort. Many remote LLM providers are aggressively competing for market share. They want developers building on their APIs. They want enterprise applications tightly coupled to their platforms. They want AI agents, copilots, workflows, and customer experiences to depend on their models. To make that happen, pricing remains highly attractive relative to the value delivered.

That does not mean the economics of generative AI are stable. It means the market is still being shaped by investor capital, strategic pricing, and growth expectations. Providers are racing to establish position, and enterprises are benefiting from that race. But no market stays in that phase forever. At some point, investors will expect durable profitability. At some point, weaker providers will disappear, consolidate, or retreat. At some point, the survivors will have more leverage and much less reason to price primarily for adoption.

That’s when the token trap closes.

Enterprises that build deep dependence on remote models during the subsidy phase may find that what seemed inexpensive at pilot scale becomes punishing at enterprise scale. The application that costs $1,000 per month today may cost 10 or 20 times that amount a few years from now, not only because usage has increased, but also because the market has repriced the dependency.

Easy to adopt, expensive to exit

Cloud computing followed a similar path, with many enterprises mistaking short-term convenience for long-term economics. In the early years, the case was compelling and largely accurate. Move faster, reduce friction, avoid capital spending, and scale with ease. Those benefits were real. Many organizations made architectural decisions that prioritized speed over leverage. They became dependent on managed services, provider-specific tools, and operating models that were easy to adopt but expensive to unwind.

Years later, many enterprises discovered that their cloud bills were much higher than expected and their exit options much narrower than advertised. That was not because the cloud failed. Architectural dependency eventually became financial dependency.

Generative AI is repeating that pattern, only faster. The integration barrier is lower, the pressure to adopt is higher, and the pace of enterprise experimentation is far greater. As a result, companies are wiring remote LLMs into applications, workflows, and agentic systems with very little thought about how these costs will behave in the next five to 10 years.

Agentic AI makes things worse

The more enterprises move from simple prompt-response systems to agentic architectures, the more dangerous the token trap becomes. Agents are not single-call systems. They plan, deliberate, retrieve information, invoke tools, evaluate results, retry steps, and often coordinate with other agents. Each of those actions consumes tokens. Costs no longer rise in a neat linear fashion. They compound.

This matters because agentic AI is increasingly being presented as the future of enterprise automation. It’s true in many cases. But if an enterprise builds agentic systems primarily on remotely hosted intelligence, it is also building future business processes on top of someone else’s pricing model. That is a major strategic risk. The more successful those systems become, the harder they are to replace. The harder they are to replace, the more pricing power shifts to the provider.

This is how businesses end up operationally dependent on a cost structure they do not control.

The appeal of AI sovereignty

The answer is not to reject public models or pretend that external providers play no role. They clearly do. There will always be cases where renting frontier AI capabilities makes sense. But enterprises need to stop assuming that renting is the default for every workload.

AI sovereignty is the alternative that deserves much more attention. That means building, tuning, deploying, and governing models inside the enterprise for use cases where long-term control matters more than access to the absolute frontier. Enterprises need to recognize that most business applications do not need a world-class general-purpose model. They need a model that is good enough for a specific purpose, aligned to the enterprise’s data, governed by the enterprise’s rules, and operated at a predictable cost.

It’s a very different way of thinking.

A self-hosted or enterprise-controlled model may not match the rich feature set of the largest public offerings. It may lack the same breadth, polish, or marketing appeal. But for many internal business tasks, those factors do not matter.

Here’s the most critical question to guide your architectural direction: Can a sovereign AI model solve the problem reliably, securely, and economically over time? If the answer is yes, owning that capability may be far more strategic than forever renting something with more power than you need. In effect, the enterprise becomes its own provider for the workloads that matter most.

Prepare for changing markets

Too many companies still treat generative AI architecture as a tactical IT issue. It is not. These decisions directly affect cost structure, operating flexibility, data control, and long-term competitiveness. If AI becomes a force multiplier across the business, the economics of AI become strategic to the business itself.

The companies that get this right will not necessarily be the fastest adopters. They will understand the difference between experimentation and dependency. They will use external models when it makes sense, but they will also invest in sovereign capabilities where ownership matters. They will think like architects, not consumers.

Here’s the takeaway: Cheap tokens come with strings. They are a gateway to a dependency model that will typically look very different once providers stop pricing for growth and start pricing for leverage. Enterprises cannot keep mistaking today’s bargain for tomorrow’s reality. Boards and executive teams need to act now to get ahead of this issue. The key question is not whether generative AI creates value. It clearly does. The real question is whether the enterprise can still afford and control the value it creates once the market matures.

Agentic AI has moved from conference hype to a budget line item. This is where the conversation gets more interesting and more uncomfortable. Unlike traditional AI systems that respond to a single prompt, classify a document, recommend an action, or generate a summary, agentic AI systems are designed to pursue goals. They plan, call tools, inspect results, retry failed steps, consult memory, hand off tasks to other agents, and sometimes critique their own work before producing an answer or taking an action.

That extra autonomy is the value proposition. It also introduces the cost problem.

A single chatbot interaction may consume a few thousand tokens. A useful agentic workflow can consume hundreds of thousands or millions of tokens per day because it does more than answer a question. It decomposes the problem, retrieves context, reasons through options, invokes APIs, checks the output, and often runs multiple passes before reaching a result. Therefore, the economics need to be understood at the level of “agent instances,” not just model calls.

For the estimates below, I am using a blended token cost of $3 dollars per million tokens. This is not intended to reflect a single vendor’s list price. It is a blended planning figure that assumes a mix of input and output tokens, reasoning steps, retrieval-augmented generation, summarization, tool calls, memory updates, and occasional use of larger context windows. Some enterprises will pay less through volume discounts or by routing work to smaller models. Others will pay more by using premium models, long-context prompts, web browsing, large document ingestion, and repeated reasoning loops.

The basic formula is straightforward. If an agent consumes 2 million tokens per day, it consumes 730 million tokens per year. At $3 per million tokens, that single agent costs about $2,190 per year in token burn. That number sounds surprisingly low until you multiply it by the number of agents, workflows, and users, plus the surrounding infrastructure required to run these systems safely.

What an agent really costs

In the model used here, the annual token-only cost per agent ranges from about $1,095 to $3,833, depending on the use case.

• A lightweight HR recruiting or onboarding agent that consumes one million tokens per day costs about $1,095 per year.
• A more demanding software engineering agent that consumes 3.5 million tokens per day costs about $3,833 per year.
• Customer support agents are roughly $2,190 per year.
• Legal contract agents cost about $2,409 per year.
• Security triage agents run about $2,738 per year.
• Research agents are around $3,066 per year.

These figures are useful but incomplete. They include only LLM token consumption and exclude orchestration platforms, vector databases, observability, model evaluation, security controls, workflow monitoring, human review, enterprise application integration, data pipelines, audit logging, prompt management, and the engineers needed to build and maintain the systems. In real deployments, I would expect the all-in operating cost to be two to five times the raw token cost. For regulated or mission-critical environments, the multiplier can be even higher.

This is where many agentic AI business cases become less clear. The model call may be inexpensive, but the system around the model is not. An agent that can update a customer relationship management system, approve a refund, generate a purchase order, or recommend a security containment action needs guardrails, permissions, logging, rollback mechanisms, and human escalation paths. These are not optional features. They are the difference between a demo and an enterprise system.

Use cases that make economic sense

Customer support is one of the more obvious use cases. A typical support automation deployment may use eight agents: an intake classifier, a knowledge retrieval agent, a response drafting agent, an escalation agent, a quality review agent, a CRM update agent, a sentiment detection agent, and an analytics agent. At two million tokens per agent per day, each agent costs about $2,190 per year in token burn, bringing the annual total to roughly $17,520. If that system deflects even a modest number of tickets or improves agent productivity, the economics can be attractive.

Sales development is another practical example. A five-agent system for account research, lead enrichment, email personalization, CRM updates, and follow-up scheduling may consume 1.2 million tokens per agent per day. That results in an annual cost of about $1,314 per agent, or $6,570 for the full agent team. This can be compelling if it improves pipeline quality, but it can also be wasteful if agents generate low-quality outreach at scale. The cost of brand damage is not measured in tokens.

Software engineering is more expensive but potentially more valuable. A 12-agent system covering requirements analysis, architecture, code generation, testing, review, security checks, documentation, CI debugging, refactoring, release notes, dependency analysis, and hot-fix support may consume 3.5 million tokens per agent per day. That works out to about $3,833 per agent annually or roughly $45,990 for the full system. Compared with engineering salaries, the token cost is small. The real question is whether the system reliably improves throughput without increasing defects, security vulnerabilities, or maintenance complexity.

Security operations also fit the agentic model because the work is repetitive, time-sensitive, and context-intensive. A 10-agent security triage system could include agents for alert triage, log analysis, threat intelligence, endpoint investigation, network investigation, incident summarization, ticketing, compliance evidence, escalation, and post-mortem. At 2.5 million tokens per agent per day, the annual token cost is about $2,738 per agent or $27,375 for the system. This is easy to justify if it reduces alert fatigue and accelerates response, but risky if the agents hallucinate causality or bury critical signals in confident summaries.

Finance, legal, healthcare administration, market research, HR, and supply chain are also viable:

• A six-agent finance close system may cost about $9,855 per year in token burn.
• A four-agent legal contract review system may cost about $9,636.
• A seven-agent healthcare administrative workflow may cost about $13,797.
• Competitive intelligence using six agents may cost $18,396.
• HR recruiting and onboarding using five agents may cost $5,475.
• Supply chain planning and exception management using eight agents may cost $21,024.

Across all 10 example use cases I’ve mentioned in this section, the model assumes 71 agents and a total annual token burn of about $175,638.

The comparison with traditional AI

The economics of agentic AI should always be compared with simpler approaches. Traditional AI, workflow automation, rule engines, robotic process automation, and non-agentic LLM calls are often cheaper, easier to govern, and more predictable. Agentic AI is usually overkill for tasks like classification, extraction, summarization, routing, or drafting within a narrow context. A deterministic workflow with a single model call can do the job at a fraction of the cost and risk.

Agentic systems make sense when the process requires judgment across multiple steps, dynamic planning, tool use, exception handling, and adaptation to incomplete information. They are valuable when the path to the answer cannot be fully scripted in advance. They are less valuable when enterprises use them as a trendy substitute for basic automation.

The best architecture is usually hybrid. Use traditional automation where the process is stable. Use non-agentic AI where the task is bounded. Use agentic AI only where autonomy creates measurable leverage. That means fewer agents, tighter scopes, explicit budgets, model routing, token monitoring, and human checkpoints for high-impact decisions.

The financial mistake many organizations will make is to treat agents as digital employees with near-zero marginal cost. They are not. They are probabilistic software components that consume tokens, trigger tools, create operational dependencies, and require supervision. The token bill may be manageable. The governance bill may not be.

Agentic AI can absolutely be worth the money. In many cases, the annual token burn for a useful agent team is less than the loaded cost of a single employee. But that is not the same as saying it is cheap. Companies must measure agent cost per completed business outcome, not cost per prompt or cost per model call. In the end, the question is not how much an agent costs. Ask yourself the right question: Does the autonomy an agent provides outweigh the complexity it introduces?