(Website-friendly American English, consistent financial/tech jargon, retains original narrative tension & paragraph structure; all company/model names keep official English branding)

Start with one staggering number: $800 billion.

This is the total AI capital expenditure the five U.S. tech giants—Google, Amazon, Meta, Microsoft, and Oracle—plan to pour into artificial intelligence in 2026.

What does an $800 billion outlay actually mean? The full U.S. national defense budget for 2026 lands right in this same range. Put plainly: the pace at which five private corporations burn cash on AI has matched the military spending of the world’s most powerful armed forces.

Worse still, this is merely the opening chapter. Forecasts from The Kobeissi Letter project this figure will skyrocket to $1.1 trillion by 2027. AI capital spending will account for 3.2% of U.S. GDP—surpassing defense spending (2.7%) for the first time in history.

Emphasis on the phrasing: for the first time in history. A single corporation’s AI budget now exceeds the entire military expenditure of an entire sovereign nation. This is not a sci-fi movie plot; it comes straight from a recent Goldman Sachs research report.

Mark Zuckerberg confirmed Meta will allocate $125–$145 billion to AI this year, a sum nearly equivalent to the annual GDP of a small-to-midsize country. Amazon’s commitment is even more extreme at $200 billion, followed by Microsoft at $190 billion and Google at $180 billion.

Context matters: last year, their combined AI investment stood at only $410 billion. In just 12 months, total spending has nearly doubled.

You may ask: Have they lost their minds?

The short answer: They sense an existential sense of urgency. Global AI competition entered an all-out, cutthroat phase in 2026. Qualcomm acquired AI infrastructure firm Modular for $3.9 billion; onsemi bought edge AI developer Synaptics for $7 billion; Salesforce shelled out $3.6 billion in cash to absorb intelligent customer agent platform Fin. These takeover deals represent just the tip of the iceberg.

The real battlefield lies in foundational infrastructure buildout. Goldman Sachs’ latest report predicts global AI capital expenditure targeting computing hardware, data centers and power infrastructure will hit roughly $7.6 trillion between 2026 and 2031—note the unit: trillions, not billions.

Annual investment will surge from this year’s $765 billion all the way to $1.64 trillion by 2031.

Where is all this capital flowing? Semiconductor chips, hyperscale data centers, and power grid infrastructure. To simplify: the United States is rebuilding a digital-age power grid with funding levels comparable to national defense mobilization.

They are no longer gambling on whether AI will succeed. They are betting everything on one critical premise: if AI reshapes every industry worldwide, they cannot afford to end up on the losing side.

This is the power of FOMO—fear of missing out—amplified to a trillion-dollar scale.

Every story has a counterpoint.

Just last month, the collective market value of the Magnificent Seven erased $2.3 trillion. Investors are growing increasingly anxious: you’ve burned through trillions—where are the profits?

How many active daily users rely on Microsoft Copilot? Has Amazon’s AI commerce assistant boosted seller revenue? When will Google’s Gemini finally turn consistent profits?

NVIDIA revealed delays exceeding 12 months (pushed to 2028) for its Kyber NVL144 server racks, caused by flawed PCB manufacturing processes. News of the setback triggered a broad selloff across AI hardware stocks. Yangtze Optical Fibre & Cable plunged 14.8% in a single trading day, while Kingboard Laminates fell 12.6%.

Large language model startups face equally tough headwinds. Zhipu AI’s stock dropped 14.7%, while both Doubao and Qwen removed user-customizable agent functionality for their platforms.

Capital markets are now confronting a brutal new question: the industry’s focus has shifted from “who built the most powerful model” to “who can actually generate sustainable revenue.”

This does not signal an AI bubble burst. Instead, artificial intelligence has entered its value realization phase. Any firm surviving solely on pitch decks with zero real income will be ruthlessly eliminated by market forces.

Against this backdrop of $800 billion in annual U.S. AI investment, what moves are Chinese tech companies making?

They are pouring in resources at full tilt as well.

Only yesterday, Tencent officially launched Hunyuan Hy3, its new large language model. Boasting 295 billion total parameters with 21 billion activated parameters, the model is fully open-sourced under the Apache 2.0 license, priced as low as $0.15 per million tokens.

Two core pillars define Tencent’s strategy: open-source access and ultra-low pricing. The company aims to eliminate computing power shortages and price barriers for AI adoption. Hunyuan Hy3 has already been integrated into nearly 50 internal business verticals including WorkBuddy, CodeBuddy, Yuanbao and ima, driving a 20x jump in daily token consumption.

Elsewhere, Kuaishou’s Lingting AI video platform closed a RMB 13.824 billion (approximately $2 billion) financing round, with both Alibaba and Tencent participating as investors. Starfan Intelligence secured over RMB 300 million in funding to accelerate real-world AI deployment.

One unignorable reality remains: the $800 billion the U.S. will spend on AI in 2026 dwarfs China’s entire cross-industry AI investment, which may not even reach a fraction of that figure.

The gap is widening—and expanding at an exponential rate.

Many argue the AI arms race is a game exclusive to tech giants, irrelevant to ordinary people.

This could not be further from the truth.

When five major tech firms’ AI budgets surpass the entire U.S. defense budget, AI ceases to be merely a niche tech vertical. It evolves into national strategic infrastructure. It will determine three pivotal outcomes over the next decade: Who sets the global standards for computing power? Who controls the backbone AI infrastructure? Which nations seize the upper hand in the next industrial revolution?

Chinese internet companies have long operated with a core mindset: follow global leaders and iterate via incremental micro-innovation. But within AI, the window for late followers is rapidly slamming shut.

When your rival deploys funding matching an entire country’s defense budget, evaluating investment purely through a startup’s narrow ROI lens means you have already lost from the start.

Tencent’s Hunyuan Hy3 is open-sourced with shockingly affordable pricing. Huawei’s Ascend chips close the performance gap every quarter. ByteDance’s Doubao serves hundreds of millions of daily users. Baidu’s Ernie Bot deepens industrial integration across countless sectors.

Yet there are no shortcuts to bridge the divide between $800 billion and $1.1 trillion in competing investment.

China’s domestic AI industry has no room to retreat.

As you read this article tonight, thousands upon thousands of GPUs are powering up inside data centers across the Atlantic. Engineers there are advancing at a pace we can barely fathom, racing ahead without pause.