Gartner has slammed AI vendors’ lack of transparency, saying developers are facing sharply increased costs from coding agents.

Since the main AI coding agent vendors have shifted from seat-based licensing to consumption-based pricing, developer teams now face highly variable cost structures.

Developer teams face an emerging problem of escalating costs, said Nitish Tyagi, senior principal analyst at Gartner. AI coding bills were leaping from $20 or $100 to $2,000 to $5,000 per developer per month, while in extreme cases, the bill might hit $20,000 in token charges.

However, software engineering departments get little insight into how token consumption is calculated and billed, making it difficult for them to forecast and control costs accurately. AI coding vendors have not yet delivered built-in features to allow developers to optimise costs in AI coding agents, resulting in cost escalation, Gartner said.

“None of the vendors have incredible features when it comes to cost optimisation,” said Tyagi. Instead, he said vendors were focusing on the concept of “tokenmaxxing” to “boost the high” of token consumption, suggesting that if

developers increase the number of tokens, they will increase productivity gains.

“There is no direct relation between the increase in token consumption and an increase in productivity gains,” he said.

Gartner recommends developer teams optimise token consumption and adopt strategies such as context engineering practices, where software engineers improve the input context provided to AI systems. Another recommended strategy is model routing, where engineering and platform teams direct simpler, high-frequency tasks to smaller models, using frontier models only for complex, high-value work

“All of these things will improve the output quality, and, therefore, will increase the productivity gains as well, so while there is no direct relation between tokenmaxxing and productivity gains, there is a relation between optimisation of token consumption with the output quality,” Tyagi said.

As a result of the lack of cost optimisation tools among vendors and consumption-reduction strategies among users, AI development is in a situation where a developer’s coding agents may cost more than they earn, at least in some parts of the world. Gartner predicts that by 2028, AI coding costs will overtake the average developer’s salary due to rising LLM token consumption and the shift to consumption-based licensing models.

“We’re not saying AI token cost will be higher than every developer’s salary on the planet, because US salaries tend to be higher than in India, for example. But current token costs are already more than most of the salaries in India,” he said.

However, the cost of coding agents does not vary according to where they are consumed, meaning AI coding costs for a developer in India may be equivalent to the salary of an engineer with four to six years’ experience, Tyagi said.

“We’re not saying AI token costs will be higher than every developer’s salary on the planet, because US salaries tend to be higher than in India, for example,” Tyagi said, adding that token costs do not vary by location, and in India they may already be equivalent to the salary of an engineer with four to six years’ experience.

New data from SAP Concur has apparently confirmed employees are still ignoring company policies, with three in four UK business travellers using (or would use) unapproved AI tools.

Data from the company’s Global Business Travel Survey shows that, while organisations are investing heavily in AI, they’re not giving workers access to the right tools.

As a result, employees are opting to use other, more suitable alternatives and often prefer publicly available chatbots such as ChatGPT and Gemini.

Two in five (43%) said they use unapproved tools because they prefer different options to what they’re provided, but one-third (32%) say their organisation doesn’t even offer any AI tools to support planning or booking travel, forcing them to look elsewhere.

The data shows that remote workers (81%) are more likely to use shadow AI than office-based employees (71%), as are younger workers like Gen Z (79%) and Millennials (76%) compared with Gen X (62%) and Boomers (49%).

As for what employees are looking for when it comes to AI supporting their travel needs, the biggest use case is planning itineraries (37%). Tracking expenses (30%), assessing travel risks (26%), making changes to their trip (24%) and completing expense reports (21%) are also common use cases for the productivity-boosting technology.

But business leaders already know the scale of the problem – 100% of CFOs are concerned about shadow AI in business travel, the report found.

“Leaders must educate workers on the risks and provide T&E tools that deliver the desired level of AI support,” EMEA VP for Concur Travel Paul Dear noted.

Looking ahead, the fix must come from organisations – workers clearly aren’t too keen on having to navigate to third-party tools, and are doing it out of necessity. More than a third actually want their AI tools integrated into their productivity software (38%), communications platforms (36%) and CRM systems (32%).

Flexera has published research showing that every organisation surveyed uses generative AI public cloud services. It also found that 85% now see managing cloud costs as their main cloud challenge.

The findings point to growing pressure on technology budgets as companies juggle hybrid cloud estates, AI spending and limited visibility over usage. Some 17% of organisations exceeded their public cloud budgets in the past year, while estimated wasted cloud spend rose to 29%, reversing a five-year decline.

Hybrid cloud has become the dominant operating model in the survey. Flexera found that 69% of organisations now use a hybrid cloud approach, rising to 78% among those with more than 5,000 employees.

Spending patterns also point to larger, more complex estates. Among organisations spending more than USD $500,000 a month on cloud, 79% operate hybrid environments.

The report suggests AI is adding a new source of cost volatility. While all respondents said they use generative AI public cloud services in some form, 45% described that use as extensive and 30% said cost unpredictability was one of the biggest challenges in scaling AI workloads.

Chris Andersen, Chief Financial Officer at Flexera, linked those pressures to broader changes in how finance teams track technology spending.

“The conversation around cloud costs has shifted significantly. It has moved from spending more on technology to solve problems to managing increasingly complex environments that have often evolved organically over time.

“Many organisations have not intentionally designed hybrid or multi-cloud strategies. Instead, these environments emerge through acquisitions, new business requirements or teams independently adopting different platforms. As a result, finance leaders are being asked to manage technology estates that are much harder to monitor and optimise.

“The challenge is that complexity itself creates inefficiencies. The more environments organisations operate across, the harder it becomes to maintain visibility into what resources are being used, whether they are delivering value and where opportunities exist to reduce unnecessary spend,” Andersen said.

The survey also points to a more formal approach to cloud oversight. Flexera found that 71% of organisations now have a Cloud Centre of Excellence, while 63% have established dedicated FinOps teams.

As enterprises race to deploy AI across business functions, many CIOs and CTOs are finding themselves responsible for systems they may not fully oversee, creating a new governance challenge for technology leaders.

A new IBM Institute for Business Value survey of 2,000 technology executives found that two-thirds of CIOs and CTOs are being held accountable for AI systems they do not fully control. The survey also found that 70% of respondents said technology is being deployed across the business faster than IT can track it, while 77% said AI adoption is already outpacing governance capabilities.

Surveyed executives expect the number of AI agents deployed within their organisations to increase by 38% by 2027, yet only 11% said they are fully prepared for the scale of AI-agent deployment they expect over the next year. According to IBM, 80% of respondents also reported CEO-driven mandates to accelerate AI transformation efforts.

“For CIOs and CTOs, the challenge now is scaling AI systems that operate continuously and autonomously, often within governance models and architectures designed for a far slower, more predictable environment,” Matt Lyteson, CIO at IBM, said in a statement accompanying the report.

Zendesk has rethought service pricing in an AI-first era beyond the evolution from seat-based to token-based pricing – the company now commits to outcome-based pricing marking one of the most important commercial shifts in the enterprise AI market.

Announced during the company’s annual Relate conference, the new model charges customers only when its AI systems have successfully revolved support interactions.

The company said every resolution would be independently verified with its own dedicated AI evaluation model, meaning that low-value exchanges would be excluded from billing.

The pioneering move reflects growing pressure across the AI industry to prove tangible business value rather than simply showcasing model capabilities for high and often wasted costs. With organisations now becoming more cautious about AI spend, vendors are increasingly being asked to demonstrate measurable ROI instead of charging for access alone, and Zendesk’s shift in pricing model holds the company accountable to pass on ROI to customers.

However, what this means in terms of revenue is less clear. For years, SaaS companies have largely relied on predictable subscription pricing tied to the number of users or licenses.

Zendesk’s approach effectively treats AI agents as digital workers whose compensation depends on measurable outcome, rather than availability. For Zendesk’s customers, it means they’ll only pay when meaningful work is completed.

As Shashi Upadhyay, President of Product, Engineering, and AI, put it in an interview at the event: “Stop thinking of agents as software… start thinking of them as a unit of labour.”

This marks the first time that a company like Zendesk could risk incurring costs without being able to recuperate them from clients, and it marks a major turning point for revenue models in general. The implications could extend far beyond customer service, as companies are expected to experiment with pricing tied to sales conversions, resolved tickets, completed workflows, productivity improvements and other outcomes.

This new strategy also reflects the growing competition within enterprise AI – looking ahead, pricing innovation may become just as important as model performance in acquiring and retaining customers.

So while Zendesk may be among the first to implement such pricing at scale, it offers a glimpse into how AI software could likely be commercialised in the future.

Workers are struggling to understand how to best use AI tools for their daily tasks due to a widening disconnect between employers and their employees, new research has claimed.

A new study from Mitel found that AI tools are often misaligned with how employees actually work, leading to widespread frustration.

This is leading many workers to bring in their own familiar tools from outside of work, potentially opening up their organisation to security threats and other risks.

The study found 71% of workers admitting to have used non-approved communication tools for work-related purposes, increasing the risk of the likes of data exposure, compliance breaches, cybersecurity threats, and a loss of visibility and control.

A similar proportion (71%) said they felt pressured to just “make it work” with tools not designed for their needs, showing the apparent disconnect or disinterest from senior management.

Overall, while just over half (51%) of workers said they regularly use AI tools, only 37% reported feeling very comfortable using them in their day-to-day work – with 70% saying they felt their organisation does not adequately support AI use.

“Organisations are making significant investments in AI, communication infrastructure, and modernisation,” said Eric Hanson, CMO at Mitel. “Yet more than half of employees report that these tools fall short at the moments that matter most.

“The challenge is not a lack of technology, but a lack of alignment with the realities of work. In fast-moving, high-pressure, and increasingly mobile environments, communication must be immediate, reliable and context-appropriate — or it risks breaking down precisely when it is needed most.”

“Closing the experience gap is now essential to reclaiming control and ensuring that communication investments translate into real-world performance and impact.”

More than 20,000 SAP customers are “stuck” on legacy ECC systems due to customisations, according to the CEO of SAP partner MyWave, Geraldine McBride, and many won’t migrate anytime soon. At Sapphire 2026, SAP offered them a path to AI, but only if they commit half their maintenance spend to the cloud first.

The offer, confirmed by SAP Chief Strategy Officer Sebastian Steinhaeuser during a media briefing, enables SAP to extend limited AI capabilities to customers still running its software on-premises during their transition to cloud ERP. But it comes with a significant condition: Customers must shift at least 50% of their maintenance spending to the cloud before they can enable Joule assistants on premises.

It’s a another small crack in SAP’s resolve: For years, it has been upsetting customers using its legacy systems by telling them the only way to access its latest innovations is to migrate to its latest platform, S/4HANA, in the cloud.

“There’s no confusion at all” about the latest change, said Steinhaeuser. “To be super clear, our Joule assistants and agents are designed for cloud. This is destiny and destiny is unchanged. […] However, we also want to meet customers where they are. We’ve put in a lot of work technically to make Joule available to ECC and S/4 customers who have committed to their journey.”

An SAP spokesperson confirmed the 50% requirement and said the company has no plans to extend the ECC end-of-service deadline beyond 2027, with extended support available through 2030 at additional cost.

For customers already on SAP’s Cloud ERP Private offering, the deal is sweeter. SAP will activate up to three Joule Assistants at no additional cost to cover common workflows in finance, HR, and the supply chain. New customers will be guided through activation during onboarding; existing customers can contact their SAP representative.