Tertius News is an AI-native newsroom: an AI model reads the linked source articles below and extracts what each outlet reported, so you can compare their coverage side by side. How this works →

Expensive AI binge leaves companies balking at soaring bills

As AI companies shift from subsidized pricing to profit-driven models, costs are rising sharply, driven by expensive AI agents. Enterprises are responding by switching to open-source or smaller specialized models, signaling a potential slowdown in AI adoption across industries.

By Tertius News AI Desk 2 outlets · 3 articles Version 1 Coverage May 31, 2026 Published Aug 13

Lead

Artificial intelligence is becoming increasingly expensive, prompting companies worldwide to reassess their deployment of the technology. After years of deliberately low pricing that fueled rapid adoption following ChatGPT's debut, AI providers are now raising costs—and businesses are feeling the pinch. Reports from multiple outlets indicate that the era of cheap AI is ending, with major players like OpenAI and Anthropic preparing for public offerings that require sustained revenue generation.

Coverage Comparison

Coverage of this trend has been consistent across major English-language media, including Dawn (a Pakistani daily) and the South China Morning Post, both of which emphasize the end of what industry insiders call "subsidised intelligence." While the core narrative—that AI costs are rising and companies are adapting—is universally reported, some outlets go further in detailing specific mitigation strategies. For instance, a report by Dawn highlights the growing shift toward free, open-source models and smaller, specialized alternatives, whereas the South China Morning Post focuses primarily on the structural reasons behind price increases, such as AI agents and infrastructure constraints.

Key Claims

The "Subsidised Intelligence" Era is Ending: According to Kevin Simback of startup incubator Delphi Labs, quoted in reports from Dawn and the South China Morning Post, investors subsidized AI usage to build market share after ChatGPT's launch. Now, as AI leaders like OpenAI and Anthropic pursue public listings, they must prioritize profitability, marking a fundamental shift in pricing strategy.

AI Agents Drive Up Costs: Multiple sources, including Dawn and SCMP, attribute rising prices to the emergence of AI agents—software tools that execute tasks autonomously. Unlike chatbots, agents can spawn dozens of parallel processes, each consuming expensive computational tokens. This can multiply token usage by dozens of times compared to simple queries, as stated in the reports.

Infrastructure Shortages Compound Costs: Both Dawn and SCMP note that computer chips and data centers cannot keep pace with AI demand, creating supply constraints that add further price pressure and uncertainty to the industry.

Companies Switch to Cheaper Models: According to a single report from Dawn, some businesses are responding by adopting free, open-source AI models or smaller, specialized ones. The report cites "dramatic" price differences, with smaller models costing as little as five cents per million tokens compared to $15 per million for larger ones. This claim has not been independently verified by other outlets in this coverage.

Overuse Leads to Tokenmaxxing: A report from Dawn quotes analyst Jack Gold of J.Gold Associates describing a phenomenon called "tokenmaxxing," where overzealous AI adoption results in token costs exceeding employee wages within months. Mark Barton of consultancy Omniux also notes that developer-focused AI costs have "grown exponentially."

Perspectives

AI Industry and Investors: Companies like OpenAI and Anthropic are transitioning from loss-leading pricing to sustainable models, which requires raising prices. Their imminent IPOs demand profitability, making price increases necessary for long-term viability. This perspective frames rising costs as a natural market maturation.

Enterprise Customers: Businesses that adopted AI based on initial low prices are now reassessing ROI. Cost overruns, especially for coding and operations, have led some to curtail usage or switch to alternative solutions—signaling a potential backlash if price increases continue unchecked.

Analysts and Consultants: Experts quoted in the reports caution that while AI offers value, indiscriminate usage is unsustainable. They advocate for strategic adoption and warn that "tokenmaxxing" could erode the financial benefits of AI, even as costs fall for smaller models.

How each outlet told it

Dawn as of May 31, 2026, 12:01 PM UTC

Framing: AI is getting expensive — Neutral

Facts Included:

AI companies charged low prices to hook customers after ChatGPT burst onto the scene
The era of "subsidised intelligence" is ending
Big AI companies need to make money now
Prices are rising due to AI agents
AI agents are expensive to run
Companies are switching to free, open-source AI models or smaller, specialized models to cut costs
The price difference between large and small models can be dramatic

Read the source article →

Dawn as of May 31, 2026, 12:31 PM UTC

Framing: AI is getting expensive — Neutral

Facts Included:

AI companies charged low prices to hook customers after ChatGPT burst onto the scene
The era of "subsidised intelligence" is ending
Big AI companies need to make money now
Prices are rising due to AI agents
AI agents are expensive to run
Companies are switching to free, open-source AI models or smaller, specialized models to cut costs
The price difference between large and small models can be dramatic

Read the source article →

South China Morning Post as of May 31, 2026, 3:31 AM UTC

Framing: AI is getting expensive — Neutral

Facts Included:

AI companies charged low prices to hook customers after ChatGPT burst onto the scene
The era of "subsidised intelligence" is ending
Big AI companies need to make money now
Prices are rising due to AI agents
AI agents are expensive to run

Read the source article →

What was claimed, and who backed it

Each row is one claim, attributed to the outlet whose wording states it most clearly. Confidence rates how directly the source text states the claim — explicit and unhedged rates high; hedged, pieced-together, or internally inconsistent statements rate lower. It does not measure whether the claim is true. Status counts the distinct outlets we found asserting it — so a single-source claim can still show high confidence, and a multi-source claim can show medium. Every one of those outlets is named beside the status, so you can check the count against the list. For claims extracted before we began storing that list, the row says so: it names the outlet the claim is quoted from and states that we have not recorded which outlets backed it. Outlets wrote at different times, so a figure that evolves — a casualty count, for example — can legitimately differ between rows; check the "as of" time next to each claim's source.

Claim	Confidence	Status
ClaimAI companies charged low prices to hook customers after ChatGPT burst onto the scene. This is referred to as the era of "subsidised intelligence" by Kevin Simback of startup incubator Delphi Labs.	ConfidenceHigh	StatusMulti-source quoted from Dawn as of May 31, 2026, 12:01 PM UTC the outlets behind this status were not recorded for this claim
ClaimThe era of "subsidised intelligence" is ending, and big AI companies need to make money now.	ConfidenceHigh	StatusMulti-source quoted from Dawn as of May 31, 2026, 12:01 PM UTC the outlets behind this status were not recorded for this claim
ClaimPrices are rising due to AI agents, which are expensive to run because one task can spin up dozens of agents all working at once, each racking up charges.	ConfidenceHigh	StatusMulti-source quoted from Dawn as of May 31, 2026, 12:01 PM UTC the outlets behind this status were not recorded for this claim
ClaimCompanies are switching to free, open-source AI models or smaller, specialized models to cut costs.	ConfidenceMedium	StatusSingle-source via Dawn as of May 31, 2026, 12:01 PM UTC
ClaimThe price difference between large and small models can be dramatic, with some smaller models costing as little as five cents per million tokens compared to $15 per million tokens for larger models.	ConfidenceHigh	StatusSingle-source via Dawn as of May 31, 2026, 12:01 PM UTC

Transparency & Audit Trail

2 outlets · 3 articles consulted: Dawn, South China Morning Post 5 claims extracted Version 1 Written 2026-08-13