Last week, markets simply repeated a pattern that has become increasingly familiar. They began to plunge as investors once again asked the same questions: Is the development of artificial intelligence financially sustainable? Will the massive investments in this infrastructure pay off in the foreseeable future?
AI, however, is not the only risk facing global markets. The second is the return of inflation as high oil prices, driven by US and Israeli intervention in Iran, feed through the economy. What was originally intended to be a week-long operation to topple the regime has turned into a geopolitical conundrum with no easy solution. The issue is not merely expensive oil, but above all its implications for monetary policy.
A Divided Fed and the Search for Certainty
Federal Reserve Chairman Kevin Warsh has only reinforced fears that his words and actions will not align. His demeanor appears populist in the true sense of the word: he tells everyone what they want to hear. Yet everyone knows that Donald Trump appointed him with a clear mission – to keep interest rates as low as possible.
This poses a fundamental problem. The Fed can employ hawkish rhetoric, but unless it is backed by votes and concrete action, markets will quickly stop taking it seriously. For now, Warsh is trying to present himself as a tough opponent of inflation while avoiding anything that might displease the White House. In the long run, however, it is impossible to play both roles at once. Markets know this, which is why nervousness is growing. Yields on 30-year US Treasury bonds have reached record highs.
What, then, changed market sentiment after six days of sell-offs in AI-related technology stocks – particularly shares in memory-chip manufacturers such as SK Hynix and Micron – and caused the tide suddenly to turn?
Microsoft as a Beacon for the Tech Sector
The answer is simple: the results of a single global company.
First, however, it is worth recalling the context of this earnings season. Companies in the semiconductor and technology sectors have mostly reported results that exceeded analysts’ expectations. Alphabet and Taiwan Semiconductor Manufacturing Company (TSMC) are obvious examples. The figures have been excellent, as have the outlooks. For investors, however, that has not been enough to push stock prices higher following earnings releases.
Markets have reached a point at which optimistic expectations must be surpassed by even more optimistic forecasts. This is not sustainable in the long term because estimates cannot be raised indefinitely. That is precisely the phase in which the market now finds itself. Even exceptional results no longer guarantee a rise in the share price. Something different is required, and Microsoft understood that challenge perfectly.
The company reported strong results. Revenue for the quarter ending in June reached $90bn, an increase of 18% from a year earlier. Earnings per share came to $4.81, compared with Wall Street expectations of $4.24. Growth in the Azure division reached 43%, three percentage points above forecasts.
More importantly, Microsoft maintained the cloud division’s operating margin at 40.6% despite rapidly rising depreciation costs from previous investments.
This provides the first evidence that massive spending on AI infrastructure is generating not only higher revenue, but also genuine profit.
Microsoft also surprised investors in its enterprise software segment. Revenue in the division rose by 15%, once again exceeding expectations. The number of paying users of Microsoft 365 Copilot doubled over the previous six months, from 15 million to 30 million. This remains only a small fraction of the platform’s more than 450 million users, but the trend is clear.
The biggest difference from Alphabet, however, lay not in the results but in the companies’ approach to investment. The Google parent raised its capital expenditure forecast for this year after publishing its results. Microsoft, by contrast, left its plans unchanged. It invested $145bn in the last fiscal year, including $41bn in the fourth fiscal quarter, which corresponded to the second quarter of 2026. It also signaled to investors that it would not continue raising its investment budget recklessly.
Paradoxically, this restraint had precisely the desired effect. Microsoft demonstrated that it had a clear plan and was following it. This was not merely a race to determine which company could spend the most on AI. Microsoft showed that its investments were already increasing revenue and that it could gradually absorb the associated costs.
Investors therefore welcomed not only another set of record results, but above all the company’s discipline. In an environment in which major technology companies are trying to outspend one another by hundreds of billions of dollars, the ability to decide that enough has been invested for the moment may be more valuable than another upward revision to the outlook.
The rise in Microsoft’s share price subsequently shifted sentiment across the sector. SK Hynix gained 26%, Micron 18%, Advanced Micro Devices (AMD) 13% and Intel 11%. This happened even though Microsoft has no direct influence over their businesses.
The market reaction was not solely about Microsoft. Investors interpreted its results as evidence that AI investment is not simply a bottomless money pit. Cloud infrastructure remains fully utilized, demand exceeds supply and capital expenditure is beginning to translate into revenue and profit. If the same holds true for other companies, the current wave of data center construction is creating genuine demand for chips, memory and other infrastructure.
In a single evening, Microsoft did more for the entire AI sector than dozens of optimistic presentations combined. It not only promised further growth, but also demonstrated convincingly that AI investments are already paying off. That was exactly what markets wanted to hear.
The Complex Reality of Artificial Intelligence
Amazon’s results further strengthened positive market sentiment the following day. Revenue from its Amazon Web Services (AWS) cloud division rose by 37% from a year earlier, its fastest growth rate in 18 quarters. At the same time, Amazon’s total operating profit increased by 43% to $27.5bn. The company therefore confirmed that demand for cloud and AI services remained strong and that heavy investment in data centers was beginning to produce financial returns.
Strong results from two major companies, however, do not resolve all the sector’s problems. The principal concern remains the cyclical nature of investment. Chip and infrastructure manufacturers finance their customers, which then use the money to buy the manufacturers’ own products.
The most visible example is Nvidia. The company has invested tens of billions of dollars in OpenAI and is also negotiating financial guarantees worth as much as $250bn for the construction of new infrastructure. The talks are reportedly expected to include financing for purchases of Nvidia’s own chips. The company would therefore be indirectly helping to create demand for its own products.
The results from Microsoft and Amazon have nevertheless shown that the situation is not black and white. Some AI investments are already generating higher revenue and profit, while others depend on an increasingly complex network of investments, guarantees and interconnected deals.
It is therefore impossible to say simply whether AI is an investment bubble. The issue is far more complex. In recent years, however, markets have operated almost exclusively in binary terms: either everything is going well or the entire narrative is falling apart. The question remains how they will handle a situation defined by subtle shades of gray.