The pitch was simple enough: pay for AI tools, get more done, save money on headcount. Companies bought it. Boards approved the budgets. IT departments rolled out the subscriptions. And somewhere between the demo and the invoice, things got complicated.
By early 2026, “AI costs” had become one of the less glamorous phrases in corporate America. The same technology that promised to make everything faster and leaner was arriving as line items that nobody had quite planned for. Monthly bills that looked manageable in the pilot phase had quietly scaled into serious budget problems. Finance teams who once rubber-stamped AI spending were asking harder questions. And a few of the biggest names in tech, the ones who had done the loudest cheerleading for AI adoption, were walking some of it back. Into that moment, with impeccable timing, steps Google.
The AI Costs Nobody Saw Coming
The surprise isn’t that AI is expensive. It’s that the expense is so hard to predict. AI is increasing SaaS cost volatility, and consumption-based pricing with AI add-ons makes budgets harder to predict and control. Unlike a traditional software license with a fixed annual fee, most AI tools charge by usage – by the token, by the query, by the API call. At small scale, that feels negligible. At enterprise scale, it snowballs.
According to Zylo’s 2026 SaaS Management Index, organizations spent an average of $1.2 million on AI-native apps in 2025 – a 108% year-over-year increase – and the trajectory shows no sign of reversing. That’s not a niche phenomenon. It’s the new normal for any company that went in enthusiastically on AI tools over the last couple of years.
Part of what’s driving those costs is a phenomenon researchers are calling “token maxing.” Organizations default to the most capable and expensive AI model for every task, including simple ones a cheaper model handles just as well. A $0.05 per million token model could answer a basic FAQ, but it gets routed to a $30 per million token reasoning engine instead. Nobody built the logic to tell the difference. In practice, that’s like ordering a limousine every time you need to run to the grocery store, because nobody set up the process for calling a cab.
Enterprise deployment audits consistently find that retry logic, context window management, and retrieval augmentation add 40 to 60 percent on top of the token costs most teams are actually tracking. The bill you see in the dashboard is, in other words, just part of the story.
The Companies Getting Burned
Two case studies from mid-2026 crystallized just how real the AI cost crunch had become, and they involve two of the biggest names in tech.
A report from Fortune revealed that Uber’s CTO Praveen Neppalli Naga told The Information in April that the firm had already burned through its entire 2026 AI coding tools budget in just four months – after the company had actively incentivized adoption through internal leaderboards ranking teams by AI tool usage. The incentives worked. The budget didn’t survive them. Uber’s Operations chief Andrew Macdonald lamented that there was no clear correlation between the money Uber was investing in AI and real consumer feature development, and senior engineers confirmed no link between higher token usage and a proportional increase in consumer features with real benefits for customers.
Microsoft’s situation was equally telling. Microsoft gave thousands of engineers in its Experiences and Devices division – the group covering Windows, Microsoft 365, Outlook, Teams, and Surface – access to an AI coding tool and encouraged them to experiment. The tool became popular fast, perhaps too popular. Usage costs grew harder to defend at enterprise scale, and on May 14, 2026, Microsoft began pulling back those licenses. Engineers in the affected division were given until June 30 to switch to GitHub Copilot CLI, Microsoft’s own in-house alternative.
A report from research firm Gartner found that cheaper tokens won’t translate to cheaper enterprise AI because agentic models require far more tokens per task than standard models, increased consumption can outpace falling unit costs, and AI providers won’t fully pass through lower costs to consumers – meaning inference costs are likely to push higher, not lower. The technology getting more efficient at the component level doesn’t mean the bills get smaller if you’re using exponentially more of it.
Agentic AI workflows make this significantly worse. A single user request can trigger ten or twenty model calls instead of one. When you multiply that across an organization of thousands of employees, the math turns unfriendly very quickly. The companies most aggressively deploying AI aren’t necessarily the ones getting the most value – they may just be the ones generating the largest cloud bills.
For anyone thinking seriously about how AI is already reshaping the jobs market, the cost equation matters enormously. An AI tool that costs more than the human it was meant to replace isn’t a productivity solution. It’s a more expensive version of the original problem.
Google’s Position
This is where Google’s position becomes interesting.
While OpenAI is burning through cash trying to keep up with demand for its models, and while companies like Microsoft and Uber are scaling back AI spending after budget shocks, Google is collecting revenue from every side of the table. According to Search Engine Journal, Google Search revenue alone reached $63.07 billion in Q4 2025, a 17% year-over-year increase, and Alphabet posted its first year exceeding $400 billion in annual revenue. The AI wave that is costing everyone else money is, for Google, a tide that keeps lifting all boats.
The reason comes down to structure. Google doesn’t have to sell you an AI subscription to benefit from AI adoption. It benefits every time you search. By Q1 2025, AI Overviews had reached 1.5 billion monthly users, and Search growth has since accelerated to 17%. More sophisticated queries, longer sessions, more engagement – all of it flows back through the advertising engine that has been Google’s core business for two decades.
Alphabet plans to spend between $175 billion and $185 billion on capital expenditures in 2026, nearly double its 2025 spending. That’s a staggering commitment, and it tells you something about how confident the company is in the return. It’s not spending that money because AI is risky for Google. It’s spending it because AI is the mechanism by which Google locks in the next ten years of search dominance.
At Google I/O 2026, according to eMarketer, Alphabet repositioned Gemini and Search as the “agentic” front door to the internet, rolling out a Gemini 3.5 Flash-powered AI Search box and agents working across Search, Android, and Workspace. Google is tapping its captive Search audience to keep users engaged with AI answers and generative tools, aiming to keep high-value queries on its own properties. Its existing consumer reach and logged-in data from Chrome, Gmail, and YouTube are advantages others cannot match.
That last part is worth sitting with. Every time someone asks ChatGPT a question instead of Googling it, Google loses a data point, an ad impression, and a piece of its feedback loop. Google’s entire AI strategy is designed around preventing that from happening at scale. The agentic search features – the ones that can book your flight, track a price, or pull together a research summary without sending you elsewhere – are retention tools as much as they are product features.
The Advertising Machine That AI Can’t Dislodge
The conventional fear about AI and Google has always been that chatbots would kill search. Why click through to Google if an AI assistant can just tell you the answer? It’s a reasonable theory. The data, so far, doesn’t support it.
In Q1 2026, Google’s ad revenue reached $77 billion, up 16% from a year earlier, with Google Search and other advertising generating $60 billion, up 19% year-over-year. These are not the numbers of a company watching its core business erode. These are the numbers of a company that figured out how to make AI feed the machine rather than eat it.
The metrics Google celebrated on the earnings call describe users staying on Google longer. Chief Business Officer Philipp Schindler described the new ad inventory as additive, reaching queries that were “previously challenging to monetize.” That’s the key move. Google isn’t just defending old territory; it’s building new inventory inside the AI experience itself. Users who previously bounced off the search results page without clicking anything are now staying longer, generating ad impressions that didn’t exist before.
Alphabet guided 2026 capital expenditures at $175 billion to $185 billion, up from $91.4 billion in 2025. That near-doubling of infrastructure investment, at the same moment competitors are pulling back AI spending, is a statement of intent. Google is betting that the cost crisis hitting everyone else is a temporary pain point that actually tightens its grip on the market. If AI tools become prohibitively expensive to run as standalone products, the most cost-efficient way to access AI will increasingly be through platforms that already have the infrastructure, the distribution, and the data – which is to say, Google.
Read More: More and More Americans Are Leaving California And Where They’re Headed To Next
What This Actually Means
The AI cost crisis of 2026 is real, but it isn’t distributed evenly. Companies that treated AI as a plug-and-play efficiency tool – roll it out, watch headcount drop, collect the savings – are learning that the math doesn’t work that cleanly. While some cling to the promise of an AI revolution, the cost of adoption is proving a stubborn bottleneck, and the economics of replacing or augmenting human labor with AI may be more complicated than early forecasts implied. Uber burning through a full year’s AI budget in four months is not an outlier. It’s a preview.
For Google, the situation looks almost paradoxical from the outside. The company that many analysts spent 2023 and 2024 writing worried pieces about – Google is disrupted, search is dead, the ad model is threatened – is posting some of the strongest revenue growth in its history, powered in large part by the very AI features that were supposed to unseat it. The threat turned into a tailwind. Not because Google got lucky, but because it had something its competitors couldn’t replicate: twenty years of search infrastructure, billions of logged-in users, and an ad business that generates enough cash to fund a level of AI investment that would bankrupt a standalone AI company.
That doesn’t mean the pressure is gone. The antitrust cases are ongoing. Meta is breathing down Google’s neck on advertising market share. And the AI arms race requires constant reinvestment just to stay in place. But the companies quietly benefiting most from the AI cost crunch aren’t the ones selling AI. They’re the ones who built the pipes it runs through. Google has been in that position for a long time. It’s just more visible now.
AI Disclaimer: This article was created with the assistance of AI tools and reviewed by a human editor.