Consulting firms spent the past two years encouraging employees to use artificial intelligence across their daily work. Now, as AI consumption reaches enormous levels, they are confronting a less visible challenge: how to control the cost of using the technology without undermining its productivity gains.

McKinsey is responding by giving employees greater visibility into how much AI they consume rather than imposing blanket limits on usage.

The consulting firm tracks AI consumption at the individual-user level and sends email alerts when an employee’s usage becomes unusually high, according to Debasish Patnaik, who leads QuantumBlack, McKinsey’s AI, data and analytics group in the UK.

The alert system was introduced across the firm during the summer.

“Similar to using mobile data on a work phone, we tell them this is how you could do things to make it more cost-effective for the firm,” Patnaik said.

The shift points to a broader change taking place among large companies. Early enterprise AI strategies largely focused on getting employees to experiment with generative AI and demonstrate where the technology could improve productivity. As adoption has expanded, companies are discovering that widespread use can generate a substantial and recurring infrastructure bill.

The economics are becoming more complicated as AI providers increasingly charge for consumption. Instead of paying a simple flat subscription for access, companies can incur costs based on the number of tokens an AI model processes. Tokens are the small units of text that models read and generate. That means an employee who makes thousands of AI requests, uses lengthy context, or repeatedly asks a model to process large amounts of information can generate significantly more costs than another employee performing simpler tasks.

OpenAI said in September that its most prolific users of AI coding agents were consuming more than $7,000 worth of tokens a day, illustrating how quickly costs can rise when advanced AI tools are used intensively.

McKinsey’s own consumption demonstrates the scale involved.

By May 2026, the firm was processing about five trillion AI tokens a month, according to a company blog post. Usage was highly concentrated, with roughly 10% of users accounting for about 65% of total consumption. Consultants and software engineers were among the heaviest users.

Rather than interpreting those figures as a reason to restrict access, McKinsey is using them to educate employees about the economics of AI.

“We really believe in giving autonomy to the consultants,” Patnaik said.

The objective is to encourage employees to find more efficient ways of obtaining the same result. An employee might use a smaller model for a relatively simple task, reduce unnecessary context, or avoid repeatedly sending the same information to a model. That approach is based on a simple premise: the cost of AI should be evaluated alongside the value it creates rather than treated as an expense that must be minimized regardless of the outcome.

McKinsey says its internal AI spending has not yet reached a problematic level. Much of the firm’s usage is aimed at improving individual productivity or delivering client engagements more quickly, cases where Patnaik said the benefits still outweigh the costs.

But that calculation could change as usage continues to increase.

“Usage could be more ‘egregious’ in another six months,” Patnaik said, while noting that McKinsey has already introduced additional controls beyond individual usage alerts.

One is an internal AI gateway that optimizes requests before they reach external model providers. The firm has also introduced circuit breakers that can temporarily suspend access when token consumption becomes particularly high, allowing the company to determine whether the usage is generating sufficient value.

Caching provides another way to reduce expenditure. Responses to repeated questions can be reused rather than generated again, while consumption costs can be pooled across the business instead of tying unused capacity to individual licenses.

The measures point to a broader evolution in corporate AI management. Companies are beginning to treat AI consumption more like a variable operating expense that requires monitoring, optimization, and governance.

Other consulting firms are taking similar steps.

EY has established an “AI Value Realization Office” to oversee AI spending and has deployed an “invisible” routing system behind some specialized AI tools. The system directs requests to the model considered most appropriate for a particular task.

EY told Business Insider that the routing system, combined with other governance measures, had reduced token consumption by 60% since April.

At Deloitte, the economics of AI coding tools have also become an issue. A senior software engineer at Deloitte US told Business Insider in June that changes to GitHub’s pricing model were “already wreaking havoc” on expectations for work, with developers quickly exhausting new monthly usage quotas.

The experience of consulting firms offers an early indication of a problem likely to spread across corporate America as AI moves deeper into business operations.

The initial enterprise AI question was whether companies could persuade employees to use the technology. The next question is whether companies can make widespread usage economically sustainable. The questions matter because AI costs do not necessarily rise in line with headcount. A relatively small group of heavy users can generate a disproportionate share of consumption, while increasingly capable models can also require more computing resources.

However, the situation creates a tension between controlling expenditure and preserving productivity for executives. Excessive restrictions could discourage employees from using AI for valuable tasks, while unrestricted usage could allow computational costs to grow faster than the benefits generated.

McKinsey is now taking its experience to clients.

Patnaik’s primary role at QuantumBlack focuses on delivering AI capabilities to clients rather than managing McKinsey’s internal AI use, but the firm established a formal practice this spring to advise companies on deploying AI more cost-effectively.

He said clients have increasingly recognized over the past quarter that there is a “hidden cost that we haven’t completely thought through.”

Companies therefore need to assess the competitive benefits of AI against the returns generated by that spending.

Patnaik argues that businesses should measure the cost of AI per outcome rather than simply calculating expenditure per employee. That distinction matters because reducing token consumption is not necessarily an improvement if the resulting AI system produces weaker work, takes employees longer to complete a task, or creates additional costs elsewhere.

“What you don’t want is to take costs out here, but incur costs on the other side without knowing about it,” he said.

The implication is that enterprise AI is entering a more mature phase. The priority is shifting from maximizing adoption to optimizing the relationship between AI consumption and measurable business outcomes.

For consulting firms that helped popularize corporate AI adoption, the transition is visible. The technology is no longer an experiment sitting on the edge of the organization. At companies such as McKinsey, it is already processing trillions of tokens every month.

The emerging challenge is making sure those tokens translate into enough additional productivity, revenue, or client value to justify the bill.

The rapid expansion of artificial intelligence investment is creating new risks for global financial stability as companies pour trillions of dollars into computing infrastructure, much of it increasingly financed through debt and private credit, according to Bank for International Settlements General Manager Pablo Hernández de Cos.

The scale of the AI buildout is already large enough to influence broader economic conditions, while its effects are reaching beyond technology companies into financial markets, trade, productivity and employment.

For central banks, AI does not alter their monetary policy mandates, but it makes economic conditions more difficult to assess because the technology can affect demand, supply and financial markets at the same time, Hernández de Cos said at a conference hosted by India’s central bank.

The BIS estimates that the world’s five largest technology companies will invest more than $1 trillion in AI between 2025 and 2026. Industry forecasts indicate that global AI investment could rise from about $500 billion currently to as much as $4 trillion by 2030.

“The promise of AI is real,” Hernández de Cos said, while cautioning that its long-term economic impact would depend on policy decisions, investment in skills and infrastructure, and how broadly the gains from the technology are distributed.

The size of the investment cycle is becoming a central concern for policymakers because the financing behind the AI boom is increasingly extending beyond companies’ existing earnings.

Hernández de Cos said more of the investment was being financed through debt and private credit, with much of that funding remaining “opaque and interconnected.” That combination could become a source of vulnerability if expected AI revenues and profits fail to materialize quickly enough to justify current investment levels.

The concern is not that AI investment will necessarily produce a financial crisis. Rather, the scale of spending, high expectations surrounding future returns and increasingly complex financing structures could amplify the effects of a downturn if valuations or corporate earnings fall sharply.

“I do not say that this is where the AI boom must lead,” Hernández de Cos said. “But the scale and speed of the current investment boom, and the weight of expected commercial returns, do warrant some caution.”

He compared the current investment cycle with earlier periods of rapid technological expansion, including the railway boom and the dotcom surge, when expectations surrounding new technologies helped drive substantial investment before economic realities eventually forced valuations and spending patterns to adjust.

The AI boom is also reshaping international trade.

Countries deeply integrated into the technology supply chain, including South Korea, Singapore, Malaysia and Taiwan, have benefited from stronger export prices for AI chips and related equipment. That creates another channel through which AI can influence the global economy. A surge in demand for advanced chips and data-center equipment can boost exports, industrial production and investment in economies that supply the technology, while increasing their exposure to a potential reversal in AI-related demand.

The productivity impact could be substantial if companies successfully integrate AI into their operations.

Hernández de Cos pointed to studies showing productivity gains of between 10% and 65% for specific tasks, particularly in areas such as coding, consulting and professional writing.

Current estimates suggest AI could increase total factor productivity growth by about half a percentage point a year, depending on how quickly businesses adopt the technology and how effectively workers and capital are reallocated toward more productive uses.

The development is of the essence because higher productivity can increase economic output without requiring a proportional increase in labor and capital inputs. But those gains depend on companies redesigning processes, workers acquiring new skills, and economies moving resources away from activities made less productive by technological change.

Advanced economies are expected to benefit first because they have larger service sectors and greater capacity to deploy AI. Emerging markets face more varied prospects.

Hernández de Cos said India has a “genuine opportunity” to narrow the gap, helped by its digital public infrastructure.

The employment effects present a more complicated picture.

AI can increase the productivity of workers by helping them complete tasks more quickly, but it can also automate routine cognitive work. Hernández de Cos said job losses have so far been limited, although signs of disruption are emerging in customer service, programming and administrative roles. That puts greater pressure on governments and businesses to invest in retraining and reskilling as AI adoption expands.

The distribution of the productivity gains will be crucial. Companies that successfully deploy AI may reduce costs and increase output, while workers whose tasks are automated could face weaker demand for their skills. Economies that can move displaced workers into expanding areas of activity are more likely to capture the broader benefits of the technology.

For financial markets, the risks are concentrated in another area: the gap between expectations and earnings.

Hernández de Cos warned that lofty valuations, market concentration and opaque financing structures could create vulnerabilities if corporate profits fail to meet expectations.

The concern is considered relevant because AI investment is increasingly concentrated among a relatively small number of technology companies and infrastructure suppliers. Their spending decisions can affect semiconductor manufacturers, data-center operators, power producers and other parts of the global supply chain.

A slowdown in AI investment could therefore spread beyond the technology sector.

The opposite risk also exists. If AI adoption accelerates faster than expected and companies compete aggressively for computing capacity, electricity and specialized infrastructure, investment could remain elevated for longer, increasing demand pressures in already constrained parts of the economy.

The situation has resulted in a difficult environment for central banks. AI can increase productive capacity over time, potentially reducing inflationary pressure, while the investment boom can simultaneously increase demand for labor, equipment, electricity and construction.

The result is an economy in which the traditional relationship between demand, supply and prices becomes harder to interpret.

For policymakers, the central issue is therefore not whether AI will transform the economy. The technology is already affecting investment, trade and workplace productivity. The harder concern is whether the current pace of spending is supported by sustainable economic returns.

Analysts have noted that if productivity gains spread broadly through the economy, today’s infrastructure spending could ultimately support higher growth and stronger corporate earnings. But if investment runs ahead of commercial returns, the same financial structures supporting the AI buildout could amplify losses when expectations change.

Hernández de Cos’s warning leaves room for both outcomes. The BIS is not predicting an AI-led financial crisis. It is warning that the sheer scale of the boom means policymakers can no longer treat AI as a narrow technology-sector story.

The financial system, labor market and global economy are increasingly becoming part of the AI investment cycle. That makes the sustainability of the boom a macroeconomic issue, not simply a bet on which technology company develops the most capable model.

Global investment commitments to clean hydrogen projects have surpassed $130 billion as governments increasingly position the fuel as a source of energy security and industrial competitiveness, even as high production costs and weak demand threaten the viability of projects around the world.

More than 570 projects have reached the committed stage, supporting about 6.9 million metric tons of annual clean hydrogen production capacity, according to the Hydrogen Council’s Global Hydrogen Compass 2026 report released Thursday alongside the Hydrogen Energy Ministerial Meeting in Tokyo.

About 90% of the projects are either under construction or already operational, suggesting that a large share of the industry’s investment pipeline has moved beyond early-stage proposals.

The scale of capital committed reflects the growing role governments expect hydrogen to play in reducing dependence on fossil fuels while supplying energy-intensive industries that are difficult to electrify directly. But the investment total masks a widening commercial challenge. Developers have scaled back investments and cancelled projects as the cost of producing green hydrogen remains high and potential customers have been slow to commit to long-term demand.

The result is an industry with a substantial project pipeline but an uncertain path to achieving the production and consumption volumes envisioned by policymakers.

China Leads Renewable Hydrogen Capacity

China accounts for more than half of the world’s committed renewable hydrogen production capacity, according to the Hydrogen Council, putting it well ahead of other major markets. Europe ranks second in investment, while the United States leads in low-carbon hydrogen deployment.

The geographic distribution highlights the different policy objectives driving the industry.

China’s position reflects its broader push to expand renewable energy, manufacturing and domestic industrial capacity. Europe has used hydrogen policy as part of its wider energy-transition and industrial strategy, while the United States has supported low-carbon hydrogen through measures designed to stimulate domestic production and investment.

The growing government involvement means hydrogen is increasingly being treated as more than a decarburization technology. It is also becoming part of national strategies for securing energy supplies, maintaining industrial competitiveness and developing new manufacturing industries.

The Hydrogen Council estimates that policies already in force could support around 6 million tons of annual hydrogen demand by 2030. A further 5 million tons of demand could emerge if governments fully implement existing measures, potentially taking policy-supported demand to about 11 million tons a year.

That potential demand is crucial because hydrogen projects require large amounts of capital before production begins. Developers need confidence that industrial customers will actually purchase the fuel at prices capable of supporting those investments.

Without sufficient demand, projects can struggle to secure financing even when governments offer subsidies or other incentives. This has become one of the sector’s central problems. Policy support can lower the cost of production, but it cannot by itself guarantee that customers will be willing to pay the resulting price for hydrogen.

Green hydrogen is produced by using renewable electricity to split water into hydrogen and oxygen through electrolysis. The process can deliver very low-carbon fuel when powered by renewable energy, but it remains expensive relative to established fossil-fuel-based alternatives.

That cost gap has become difficult in sectors that were once viewed as natural markets for hydrogen.

Steelmaking, heavy transport and other hard-to-electrify industries could theoretically use hydrogen to replace fossil fuels, but companies in those sectors must weigh the cost of switching against competing technologies and their existing production economics.

Long-distance transportation faces similar challenges, with hydrogen competing against battery-electric systems and other fuels depending on the application.

The result is a difficult investment equation: hydrogen developers need customers before committing billions of dollars to production capacity, while potential customers often want evidence that hydrogen will become cheaper and more reliably available before investing in equipment and infrastructure to consume it.

Investment Boom Faces A Reality Check

The $130 billion investment figure is widely seen as an indication that governments and companies remain committed to developing a global hydrogen industry. Yet the cancellations and investment pullbacks show that capital commitments alone cannot establish a functioning market.

The next phase will therefore be less about announcing projects and more about making them commercially viable. That will require lower production costs, dependable renewable power, transportation and storage infrastructure, and customers prepared to sign long-term contracts.

Hydrogen also faces competition for capital from other decarburization technologies. In applications where direct electrification is technically and economically feasible, using electricity directly can be more efficient than generating hydrogen and then converting it back into useful energy. This makes hydrogen most compelling where electrification is difficult or impractical, rather than as a universal replacement for fossil fuels.

The industry’s challenge is consequently shifting from building an investment pipeline to proving that the pipeline can generate sustainable demand and acceptable returns.

For governments, that could mean moving beyond production incentives toward policies that create predictable markets, while it means demonstrating that projects can survive without assuming that subsidies or rapidly falling technology costs will solve the economics.

The Hydrogen Council’s figures show that the clean hydrogen industry has accumulated substantial capital and political support. The cancellations and delayed investments show that it has yet to establish an equally substantial commercial market.

The gap between those two realities, analysts note, will determine whether hydrogen becomes a major pillar of the global energy system or remains a heavily subsidized fuel with a large but uneven project pipeline.