Shares of Zhipu AI, China’s flagship pure-play artificial intelligence company, exploded as much as 35% on Wednesday before closing up 31.94% after the Beijing-based startup delivered its first set of public financial results since listing in Hong Kong in January.

The eye-catching rally came despite Zhipu posting a wider loss and slightly missing revenue expectations, underscoring how investors are pricing in explosive long-term growth potential rather than near-term profitability in China’s state-backed AI race.

Revenue for 2025 climbed 132% to 724 million yuan ($99.5 million), powered by surging demand for its large language models and AI agents. The figure fell just short of the 760 million yuan consensus forecast compiled by Reuters, but the triple-digit expansion signaled that adoption is accelerating even as the company pours money into research.

The adjusted net loss widened 29.1% to 3.18 billion yuan, reflecting heavy R&D spending typical of frontier AI developers still in the heavy-investment phase.

Founded in 2019 by a team of researchers from China’s prestigious Tsinghua University, Zhipu has quickly become one of the country’s most prominent “AI tigers” — the handful of well-funded startups racing to build large language models that can stand toe-to-toe with OpenAI and Anthropic.

Its latest GLM-5 model, released in recent weeks, claims parity with leading U.S. systems on several key benchmarks, while the company has aggressively expanded its AI agent offerings and open-source tools.

One clear bright spot: a nationwide frenzy around its open-source AI agent OpenClaw has driven token usage, the basic unit measuring computing demand, to record levels. Zhipu said more than 4 million small and medium-sized enterprises and individual developers now use its products, which are available in 218 countries and regions. That reach, combined with Beijing’s full-throated push for technological self-reliance, has turned Zhipu into a bellwether for the entire Chinese AI sector.

During Tuesday’s earnings call, CEO Zhang Peng highlighted a sharp spike in computing demand since February and said the company is fast-tracking its transition to domestic Chinese chips to meet it.

The comments carried extra weight: Zhipu was placed on the U.S. Commerce Department’s Entity List in January 2025 over alleged military links, severely restricting its access to advanced American semiconductors. Like its peers, it is now racing to build a fully indigenous supply chain, aligning perfectly with national policy.

The market’s enthusiastic reaction on Wednesday also lifted rival MiniMax, another Hong Kong-listed Chinese AI startup, whose shares rose about 16%. Both companies listed in January after raising hundreds of millions, part of a broader wave of Chinese AI firms tapping public markets to fund the enormous capital requirements of model training and inference.

Zhipu’s debut as the world’s first major pure-play AI model company to go public, at least in the conventional sense, has given investors a direct way to bet on China’s AI ambitions at a time when Beijing is pouring resources into closing the technology gap with the United States. The government’s support includes preferential access to domestic chips, data resources, and policy tailwinds that have helped offset export controls.

Yet the path remains capital-intensive and uncertain. Zhipu’s losses are widening as it scales, and the company still faces the classic frontier-AI dilemma: massive upfront spending with revenue that, while growing fast, remains modest relative to the infrastructure costs.

Full profitability could be years away, and any slowdown in domestic chip performance or tightening of U.S. restrictions could complicate its roadmap.

Still, Wednesday’s surge suggests investors are willing to overlook the red ink for now. The stock’s performance reflects confidence that Zhipu, and by extension China’s AI ecosystem, is turning the corner from catch-up mode to genuine competition. With GLM-5 already claiming benchmark parity and OpenClaw driving real usage, the company is positioning itself as more than a local player. It aims to become a global force in a market where scale, data, and government backing could prove decisive.

For a sector long viewed as dominated by a handful of U.S. giants, Zhipu’s strong debut earnings and the market’s response mark a notable milestone in the shifting global AI balance.

Oracle shares rose in early trading on Wednesday after the software and cloud giant began cutting thousands of jobs in a sweeping restructuring designed to release cash for its increasingly expensive artificial intelligence infrastructure expansion.

The layoffs, which have affected employees across the United States, India, and other key markets, come at a pivotal moment for the Larry Ellison-led company as it attempts to transform itself from a legacy database powerhouse into a front-rank AI cloud infrastructure player.

Multiple reports indicate that the latest round of cuts could run into the tens of thousands, with estimates ranging from about 10,000 to as many as 30,000 positions, potentially affecting nearly a fifth of Oracle’s global workforce of 162,000.

Rather than punishing the stock, investors pushed shares higher, reflecting a view that the restructuring may help restore confidence in Oracle’s balance sheet at a time when its AI ambitions have become a major source of concern.

This is fundamentally a story about capital intensity.

Oracle is pouring unprecedented sums into data centers, GPU clusters, networking hardware, and power infrastructure to support large AI workloads for clients that include Nvidia Corporation, Meta Platforms, Inc., OpenAI, Advanced Micro Devices, Inc., and xAI.

The company has already raised roughly $45 billion to $50 billion in debt and equity to finance this expansion, a figure that has sharply increased scrutiny over leverage, free cash flow, and the pace at which these investments can begin generating returns.

Oracle’s enormous commitment to AI-related cloud contracts, particularly its multibillion-dollar infrastructure arrangements linked to OpenAI and the broader Stargate ecosystem, has become a large source of investors’ anxiety.

Those deals promise significant long-term revenue, but they also require massive upfront spending. That mismatch between immediate cash outflows and deferred revenue inflows is now driving the company’s restructuring decisions.

In effect, Oracle is trying to rebalance its cost structure by shrinking recurring labor expenses while preserving the financial flexibility needed to continue building AI capacity.

Analysts have increasingly framed the layoffs as less about cyclical weakness and more about cash-flow engineering. Payroll remains one of the largest controllable operating costs for a technology company of Oracle’s scale. Cutting headcount allows management to redirect billions of dollars toward capex without relying solely on further borrowing.

This matters because Wall Street has become increasingly sensitive to debt-fueled AI bets. Earlier concerns over Oracle’s capital spending intensified after the company’s free cash flow turned sharply negative amid escalating infrastructure outlays.

The company’s stock has already lost substantial value this year, underlining skepticism about whether its aggressive AI expansion can be monetized quickly enough. The latest restructuring is therefore as much about reassuring investors as it is about funding physical infrastructure.

Across the technology sector, major hyperscalers including Alphabet Inc., Microsoft Corporation, Amazon.com, Inc., and Meta have committed hundreds of billions of dollars to AI infrastructure, from advanced chips to dedicated data-center campuses.

What distinguishes Oracle is that it is making this push from a relatively smaller cloud base compared with its larger rivals. That raises the stakes considerably.

AI capex is an extension of already dominant cloud businesses for Microsoft and Amazon, while Oracle sees it as a strategic repositioning that could redefine the company’s long-term growth trajectory. The job cuts, therefore, signal more than cost discipline. They represent a decisive shift in corporate priorities.

Human capital is being subordinated to infrastructure capital, and it is becoming a defining feature of the AI era: companies are no longer only cutting jobs because demand is weak; increasingly, they are cutting jobs to fund technology bets whose returns remain uncertain.

There is also a productivity narrative at play. Some analysts have pointed to Oracle’s lower profit per employee relative to peers, arguing that the restructuring may help improve operational efficiency ratios and support margin expansion over time.

However, Oracle’s AI strategy depends on sustained enterprise demand for high-performance cloud capacity and the ability of major clients to keep expanding their own AI spending. Against that backdrop, any slowdown in enterprise AI adoption, any delay in customer deployments, or any weakening in the broader tech investment cycle is expected to expose Oracle to elevated debt and a leaner workforce without the expected revenue uplift.

For now, the company is making a clear calculation: in the current technology cycle, the priority is not preserving headcount but securing the compute infrastructure required to compete in the next phase of enterprise AI.

While the market may be rewarding that discipline today, the longer-term outcome is not certain, as the future of the AI industry is still a debate.

China’s housing market posted a modest but symbolically important recovery in March, with new home prices rising for the first time in months, offering a tentative sign that the country’s long-troubled property sector may be beginning to find a floor after years of deep distress.

According to data from the China Index Academy, prices of new homes across 100 cities rose 0.05 per cent month-on-month in March, reversing a 0.04 per cent decline in February. The gain was driven by stronger seasonal demand in major cities and increased supply of higher-quality projects in core urban markets.

While the increase is modest in numerical terms, its significance lies in what it represents for a sector that has been at the center of China’s economic anxieties since the collapse of China Evergrande Group.

For nearly five years, China’s real estate market has been in prolonged turmoil. The crisis began in 2020 when Beijing introduced strict borrowing limits under the “three red lines” policy to curb the excessive leverage that had fueled years of debt-driven expansion among developers. That move exposed the fragility of heavily indebted firms, with Evergrande emerging as the most dramatic casualty.

Once China’s largest developer, Evergrande, amassed liabilities exceeding $300 billion, becoming the poster child of the country’s property bubble. Its liquidity crisis in 2021 triggered missed debt payments, stalled construction projects, and widespread defaults across the sector.

The severe fallout was marked by millions of homebuyers who had prepaid for apartments, left waiting for unfinished homes. Construction froze across numerous cities, confidence collapsed, and household wealth came under sustained pressure.

Because property accounts for a substantial share of Chinese household assets, the slump quickly spilled beyond real estate into broader consumption and economic confidence. That is why March’s price gain, however small, marks a significant shift.

It is the latest sign that the steepest phase of the correction may be easing, particularly in China’s top-tier cities. The China Index Academy itself stressed the importance of sustaining the momentum.

“The continuity of this recovery in April will be critical,” the firm said.

“If momentum can be maintained in major cities, it will help improve market expectations and lay a stronger foundation for stable market performance throughout the year.”

China’s property crisis has become as much a crisis of confidence as one of supply and demand. For years, falling prices discouraged purchases, with households postponing decisions in anticipation of further declines. Developers, in turn, struggled to generate presales, worsening liquidity strains.

A stabilization in prices, especially in key cities such as Shanghai, Beijing, and Shenzhen, could begin to reverse that psychology. There are early signs of this. Official data from February had already shown that price declines in major cities were slowing, even as lower-tier cities remained under heavy pressure.

However, Fitch Ratings’ Shi Lulu warned that the broader market backdrop remains fragile, underlining substantial challenges.

“Given weak employment conditions, elevated housing inventory and other fundamental challenges, overall market sentiment remains fragile,” she said.

Large inventories of unsold homes, particularly in lower-tier cities, continue to weigh on recovery prospects. This makes caution well-founded.

Many buyers are also increasingly turning to the secondary market, where prices have become more attractive, potentially diverting demand from new projects. Moreover, property investment has sharply declined from its peak contribution to GDP, falling from around 12 per cent to roughly 6 per cent over the course of the crisis.

For years, real estate was one of China’s most powerful growth engines, at one point accounting for roughly a quarter of economic activity when linked sectors such as steel, cement, appliances, and local government land sales are included.

Its downturn has complicated Beijing’s efforts to rebalance growth toward consumption and advanced manufacturing.

Although the sector is far from full recovery, the return to positive monthly price growth offers a potentially important inflection point after years of contraction and negative headlines dominated by Evergrande’s collapse and its aftershocks.

While the gain does not signal a full recovery, it does represent the clearest sign yet that China’s battered property sector may be starting to turn a corner. In effect, the market is moving from crisis management to cautious stabilization.