AI Users May Not Spend What Is Expected

[marcus]

From https://finance.yahoo.com/technolog...nlimited-ai-demand-for-granted-100000961.html
Title : The era of 'tokenmaxxing' is over — and investors shouldn't take unlimited AI demand for granted
By Hamza Shaban , Oct 2026

... a new blog post by Apollo chief economist Torsten Sløk: don't take AI demand for granted.

Ultimately, the key metric that will define the AI build-out isn't physical capacity or power, Sløk said, but the price users are willing to pay for the compute being built ...

An analysis of declining returns on invested capital for the 5 largest hyperscalers suggests invested capital grows faster than the earnings it generates. That mismatch gets at the heart of the AI bull case and gives investors questions to ask and numbers to watch ...

Companies are willing to pay a high price for compute, he said, if they are able to boost revenues, squeeze out better profit margins, or cut costs. But such results at a meaningful scale aren't showing up yet.

So far, he said, the AI payoff has been slow to materialize. That suggests more companies are reluctant to dedicate resources to pay for AI services. Those returns may actually emerge. But the uncertainty comes at a price.

Poster's Note : See Apollo's article at https://www.apollo.com/wealth/insig...n-spend-shows-compute-demand-is-not-unlimited

 

[marcus]

From https://finance.yahoo.com/technolog...-and-turning-to-cheaper-models-162837207.html
Title : Big companies are rethinking 'out of control' AI spending — and turning to cheaper models
By Ines Ferré · Senior Business Reporter , Oct 2026

IBM is calling attention to a growing trend among companies and their AI spending: "I think what we are seeing is that at least larger enterprises are becoming .. a lot more aware of what the true ROI (Return on Investment) of all of their AI spend is," IBM CEO Arvind Krishna said this week.

"They are also getting worried if some of the AI spending is completely out of control with no real return," he added ...

"Our computation is that for an average large enterprise, you can get going with maybe 2 or 3 racks [of infrastructure]," said Krishna. "It gives you the ability to run those (#) open-weight models currently, anywhere from 10% to 25% of the token cost of the frontier models."...

(#) An Open-Weight Model is an AI model whose core components are publicly released, allowing anyone to download it. This lets users run the model on their own computers, study how it works, and even modify it for their own specific needs. (from https://hai.stanford.edu/ai-definitions/what-is-an-open-weight-model)

 

[marcus]

IBM CEO : "... some of the AI spending is completely out of control with no real return"

This make me remember about my co-worker who used our center's supercomputer for his own learning benefit .

 

[marcus]

From https://www.techradar.com/pro/gartn...-productivity-and-savings-goals-from-using-ai
Title : Gartner report finds just 3% of organizations say they've hit productivity and savings goals from using AI
By Mike Moore , Oct 2026

... A report from Gartner found just 3% of organizations have hit their targets around productivity and efficiency, as well as trying to cut costs across their business using AI.

This is despite three quarters of leaders expecting AI technology to have a significant effect in their organization, with 85% set to spend more on AI in 2027, despite 38% of them already being over budget ...

Moving forward, Gartner says businesses should look to focus their AI efforts where the technology can make a tangible effect in key areas. It also advised organizations to spread AI knowledge across the business, with C-suite as important as entry-level workers when it comes to identifying new use cases and benefitting from such improvement.

“There isn’t a struggle to create capacity, that’s happening already as organizations redesign work,” .. “What organizations are struggling with is how to make that capacity usable and redirecting it to what is valuable.”...

Poster's Note : Gartner, Inc. is an American research and advisory firm focusing on business and technology topics, and has over 20,000 employees globally and operates in 90 countries and territories (https://en.wikipedia.org/wiki/Gartner)

 

[marcus]

Ramp data shows AI spending stalling ..

AI spending flatlined over late summer even as token volumes rose, a sign that price competition and savvier customers may be capping revenue for OpenAI and Anthropic .. token spend only grew 1.5% across August and September. To compare, in July, it grew 42% from the previous month..

the data suggests investors might be wrong to assume AI companies will keep growing at breakneck speed...

“The era of (AI) hypergrowth may be over ...

the slowdown isn’t due to a demand problem (as) AI volumes are up. Instead the point is the competitive nature of the marketplace.

“OpenAI and Anthropic price cuts have .. just made it cheaper to use AI,”...

Subsidizing use of expensive technology in order to lock in customers has always been a part of the Big Tech playbook ...

also pointed to the range of other open model companies from China and the US, as well as a new company called Jev — a lightweight tool for choosing between a few predetermined outputs — that is increasingly among the top performers on Ramp.

“For a lot of AI use cases, you just need a simple Yes/No,” Kharazian said. “So you take your already cheap models, you shift them over to Jev, and now your already cheap stuff gets even cheaper.”

The ability to complete workloads at a much cheaper cost is another factor driving the spending slowdown. That’s partly due to the discounts from AI companies, but also because companies are just getting better (on AI's use) ...

(https://techcrunch.com/2026/11/06/r...ing-stalling-just-as-anthropic-nears-its-ipo/ - By Rebecca Bellan , Nov 2026)

Poster's Note : Ramp Business Corporation is an American multinational financial technology company . As of June 2026, Ramp was valued at $44 billion with $1 billion in annualized revenue - https://en.wikipedia.org/wiki/Ramp_(company)

----------------------------------

... Revolut (the London-headquartered digital bank) .. replaced costly AI workloads with self-hosted, open-weight models from China's Z.ai.. it can use AI much more freely without worrying about token limits, which in turn has also improved productivity.

Revolut seems pretty au-fait with the world of AI, having previously experimented with Google's Gemini before opting to use Anthropic's Claude, but some heavy users were said to be spending thousands of dollars per month on tokens without generating the equivalent ROI..

Revolut's self-hosted option resulted in a 70% improvement in lead time productivity, and a 60% improvement in coding time productivity. He also praised Z.ai's GLM models for their data privacy and auditability...

However, with great flexibility comes great cost. While costs associated with direct model usage have been slashed, self-hosting requires significant outlay in terms of hardware and infrastructure...

with more companies ditching commercial AI from the likes of Claude in favor of self-hosting, it begs the question whether using less advanced models for mundane work and even self-hosting over renting could improve the economics in the years that come ...

(https://www.techradar.com/pro/we-we...i-coding-costs-by-cutting-out-claude-entirely - By Craig Hale , Nov 2026)