IMF chief Kristalina Georgieva told AFP in October that if investors lose confidence in future returns from AI investments, tanking stock market valuations, it could "lead to financial stability risks" worldwide.
That warning is now colliding directly with corporate performance across major markets. The core question confronting enterprise leadership is whether massive enterprise outlays on artificial intelligence are generating measurable financial returns or simply marching balance sheets toward a structural correction. As corporate statements and operating results begin to align, the absence of revenue and profit expansion from deployments shifts the discussion from technology benchmarking straight to macro stability risks.
Vanishing Margins and Revenue Traps
During a panel discussion with business leaders in Shenzhen, Baidu chief executive Robin Li outlined an urgent situation where enterprise deployment is failing to deliver basic economics. The expansion of the technology into practical use cases has arguably fallen short of expectations, leaving the fundamental business case unproven for corporate adopters.
"Can we say that companies are seeing higher revenues or increased profits after adopting AI? For most, the answer is no. If we fail to address this, the concern that AI might be a bubble could well become a reality."
Li pointed out that artificial intelligence adoption is not boosting profits for most companies, warning that an inability to translate models into cash flows turns bubble risks into reality. The operational reality behind that statement is visible on Baidu's own books. On Monday, the Beijing-based search provider posted a sixth straight drop in quarterly revenue, burdened by a domestic consumer slump dragging on advertising cash flows. Earlier, the company recorded its fifth consecutive quarterly revenue decline, dropping 4% year-on-year to 31.3 billion yuan ($4.6 billion) as operating profit slipped to 3.0 billion yuan, exposing the drag of sustained technology outlays against flagging top-line metrics (we covered this on 21.08).
Geopolitics, Hardware, and Enterprise Counterarguments
The financial reckoning unfolds as the Asia-Pacific Economic Cooperation meeting gathers global figures in Shenzhen, where political friction surrounds the compute supply chain. Earlier, data from IDC indicated that by 2025 local companies were projected to hold 41% of the AI chip market in China, accounting for 1.65 million out of 4 million shipped AI cards (we covered this on 02.04). Today, market jitters over the artificial intelligence boom have been heightened by Jensen Huang's resignation as Nvidia CEO after admitting an affair reportedly with a suspected Chinese spy. The scandal has intensified scrutiny over how Chinese enterprises access Nvidia's most powerful hardware to train and run AI systems despite U.S. export rules. On Thursday, Donald Trump and Xi Jinping are due to discuss AI safety alongside sensitive tech risks during APEC talks, which also coincide with a historic trilateral meeting between Xi, Trump, and Russian President Vladimir Putin, while Japanese Prime Minister Sanae Takaichi attends her first visit to China since taking office in October 2025.
Bhatia said Google's AI technology, including chatbot Gemini, can lower costs in various sectors, such as by improving efficiency in health care and agriculture.
Other corporate leaders push back against the bubble thesis by pointing to workflow gains. Karan Bhatia, Google's global head of government affairs and public policy, stated that the world is entering an AI-driven economy requiring a new vision of international norms, arguing that platforms like Gemini lower operational expenses across sectors. Brendan Nelson, president of aviation giant Boeing Global, similarly observed that a recent explosion in e-commerce and AI has driven growth in air cargo volumes. Yet operational efficiency gains have not reversed the wider financial mismatch: commercial deployments continue to demand extreme hardware budgets while failing to generate documented enterprise profit lines. Technology spending that fails to generate cash flow remains a capital sink, and without demonstrable balance-sheet returns, the current infrastructure expansion cannot sustain its current valuations.
What’s behind it
Baidu's operating profit sliding to 3.0 billion yuan exposes a harsh corporate trade-off, where massive spending on computing infrastructure fails to generate matching cash inflows. While some executives defend the technology by pointing to workflow productivity, others openly concede that practical enterprise deployments are not yielding bottom-line profits. Pouring capital into model integration without incremental returns leaves corporate balance sheets directly exposed to structural correction. The failure to convert generative systems into tangible cash flows turns heavy compute investments into a direct systemic risk.
Our stories: Baidu Revenue Falls 4% Amid Heavy Ernie AI CapEx Costs (August 21).