Wednesday, November 18, 2026

Top 5 This Week

Related Posts

AI in 2027: companies facing the real cost of uses | Brief AI


Gartner anticipates $99 billion in global spending on AI platforms and models in 2027. However, unit prices are falling among several suppliers. For companies, trade-offs also concern operating resources, the distribution of tasks and the supervision of agents.

Buyers could negotiate with an open alternative

Manish Jain, CEO of Strategic Horizon Research, does not anticipate a radical migration to open-weight models in 2027. He does, however, give them a role in negotiations: a company has another option when its proprietary supplier raises its prices or revises its terms. The extent of task transfers to these alternatives could weigh on the pricing latitude of proprietary offers.

Michael Heinrich, CEO of 0G Labs, describes a distribution in which open models absorb repetitive and large operations, while difficult reasoning remains left to proprietary models. The latter can, according to him, be billed 10 to 30 times more per token. He also observes architectures where a state-of-the-art model sets the plan and then less expensive models execute many calls. This combination could gain ground in 2027, without removing infrastructure and operating expenses: Heinrich recalls that hardware resources and engineering are paid even when the GPUs are not working.

Jain explains the continued preference for proprietary offerings by their performance, ease of use and operational support. Making open alternatives work with reliability, security and economic control requires, according to him, specialists. He expects their progress first in assistance based on internal knowledge and document processing, where data control, sovereignty and budgetary visibility take precedence over small performance gaps.

Commercial responses remain forecasts. Heinrich expects American suppliers of closed models to offer smaller, less expensive models, open-weight variations as well as more services bundled into company contracts. Jain instead envisions increased competition on reliability, safety, tooling and service commitments.

Gartner predicts a 54.2% increase in spending

Gartner’s projections are for AI platforms and models to grow from $64 billion to $99 billion in 2027, an increase of 54.2%. Nick Patience, analyst at Futurum Group, also forecasts an overall budget increase despite a cheaper token, driven by consumption. The extension of the uses of models and agents could thus prevail over unit economies.

In 2026, models have reached new cost levels. Some suppliers and customers then encouraged their employees to consume as many tokens as possible, under the name tokenmaxxing; this practice has since been largely abandoned. The use of open and less expensive Chinese models has at the same time fueled price competition between Anthropic, Google and OpenAI. At OpenAI, GPT-6 costs half as much as GPT-5.6; Google has also halved the price of Gemini 3.7 Flash compared to its predecessor.

For Forrester analyst William McKeon-White, budgeting is about more than models: companies also need to keep track of their SaaS platforms. It foresees easier management in 2027 with better tools, frameworks and systems, as well as increased budgetary attention and control. This improvement does not constitute a forecast of reduced spending.

Patience expects increased attention to the price of the completed task rather than just the volume of tokens. His example is that of a cheap model having to start again three or four times: it can ultimately cost more than an expensive offer successful on its first attempt. The quality of execution therefore enters into the arbitration he describes.

The evening letterFree · 7 p.m.

AI transforming business?

AI strategies, movements and tricks deciphered, every evening in 5 minutes. Free.

Included upon registration: our selection of the best AI guides & comparisons.

Choose your pace

Free · No spam · Unsubscribe in 1 click

Companies will have to choose who divides the tasks

Deepak Seth, an analyst at Gartner, anticipates a shift in competition toward platforms that can assign each task to the appropriate model. With more agents, businesses should coordinate multiple models instead of depending on a single offering. One possible configuration would reserve Anthropic’s Opus 5.5 for complex jobs and Haiku, less expensive, for simple requests. The choice also concerns the operator of this routing: the company itself or a service provider.

Another trade-off expected in 2027 concerns the distribution of computing between cloud and local installations to contain costs. Nick Patience observes a growing interest in on-premises deployments, even among hyperscalers. These installations are also planned to place the models as close as possible to business data.

Supervision constitutes the other project of 2027 with the expenses: agents have regularly left the controlled frameworks provided for them, causing concerns about the dangers of these systems. Seth considers their management critical as they multiply in IT. Uses should evolve from time savings on low-risk applications to measurable results or business transformations, which could go as far as the creation of applications by users or the elimination of entire departments.

Patience predicts that labs will give more weight to agents’ operational capabilities than their ease in conversation. The use of tools, the control of a computer and the reliability of a succession of steps are among the priorities he expects.

Futurum measures progress in open models

The adoption of open models is progressing, but proprietary suppliers retain a significantly higher share of measured consumption. Futurum Group published a survey of 200 companies in October: 42% said they were using open models in production in September, compared to 31% in July. Among user companies, the proportion of tokens coming from these models is expected to increase from 34% to 41% over one year.

The distribution of expenses remains different. Futurum results attribute 5% of purchases of usage-based AI services to open models. Microsoft 365 Copilot represents 36%, and Claude, at Anthropic, 29%.

The offer notably pits Cohere and Mistral against Chinese developers, including Qwen from Alibaba and DeepSeek. Mistral, based in Paris, made the weights of Mistral Large 4, also called Le Chonk, available on October 29. This multimodal model has 1,000 billion parameters, including 52 billion active, and targets programming as well as cybersecurity. The company claims better results than competing open offerings in programming tests.

Interact Analysis expects around 20% more cobots

Samantha Mou, analyst at Interact Analysis, forecasts an increase of around 20% in cobot shipments for 2027, continuing the strong growth in 2026. She expects growth to be greater than that of traditional industrial robots. These collaborative machines are designed to work safely with humans, particularly in manufacturing or assembly plants. The analyst cites AI integration, which improves performance and expands applications, as well as adoption by SMBs seeking greater efficiency as drivers.

The expected progression of robotics and physical AI in 2027 would remain geographically uneven, with a concentration in Asia-Pacific and a predominance of factory and warehouse automation. Yueqin Shen, an analyst at Omdia, a division of Informa TechTarget, anticipates a transformation in warehousing and logistics based on a survey of industrial users. It places the strongest growth in projects in autonomous guided vehicles, or AGVs, and autonomous mobile robots, or AMRs. Among respondents, more than half plan new installations within 18 months.

These robots provide goods transportation and handling within warehouses, distribution centers and factories. They can also carry out patrols around industrial installations, particularly mining. For machine vision products for logistics, Omdia projects the highest revenue growth rate, at 14.5% between 2025 and 2030. The firm largely links this outlook to the diffusion of AGVs and AMRs.

Shredder Smith
Shredder Smith
Shredder Smith is the lead curator and digital persona behind topaitools4you.com, an AI directory dedicated to "shredding" through industry hype to identify high-utility software for everyday users. Smith positions himself as a blunt, no-nonsense reviewer who vets thousands of emerging applications to filter out overpriced "wrappers" in favor of tools that offer genuine ROI and practical productivity. The site serves as a watchdog for the AI gold rush, providing categorized rankings and transparent reviews designed to help small businesses and creators navigate the crowded tech landscape without wasting money on low-value tools.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Popular Articles