Meta Platforms raised its 2026 capital expenditure forecast to a range of $130 billion to $145 billion on July 29, narrowing the guidance window while lifting the floor by $5 billion from earlier projections. The revision came alongside second-quarter earnings that showed AI infrastructure spending consuming nearly all operating cash flow—a stunning revelation that sent shares tumbling almost 10 percent in the following session despite record revenue of $60.8 billion.

The capex announcement from a single hyperscaler sets a new benchmark for AI infrastructure spending and signals sustained demand for power, land, construction, and hardware across the entire technology supply chain. Meta's second-half spending implies a run rate of $81 billion to $96 billion for the remainder of 2026, representing a 65 percent to 95 percent increase over the first-half pace of $49.1 billion.

Investors did not celebrate the ambition. Free cash flow collapsed to just $784 million in the second quarter, down from multi-billion-dollar quarterly averages in prior years, as Meta funneled every available dollar into data centers, custom silicon, and the compute infrastructure required to train and deploy its Llama 4 foundation model. The stock's 10 percent drop reflected Wall Street's growing anxiety that Big Tech's AI spending spree has crossed from strategic investment into reckless capital allocation.

The Hyperscaler Arms Race

Meta's increased guidance arrived in the context of an industry-wide capex explosion. Amazon also raised its 2026 forecast by $20 billion earlier this month, citing the cost of high-bandwidth memory required for AI servers. Data Center Digest reported that when combined with Microsoft and Alphabet, the four largest hyperscalers are now on track to spend approximately $725 billion in 2026—up 77 percent from roughly $410 billion in 2025—with analysts projecting over $1 trillion in 2027.

Nearly all of this spending targets AI infrastructure: GPU clusters for training large language models, custom silicon such as Google's TPUs and Meta's MTIA chips, and the physical data center construction required to house and power these systems. The scale represents a fundamental reshaping of global capital flows, with technology companies redirecting profit that would historically have returned to shareholders into an arms race for AI dominance.

Meta's capex intensity on a dollar basis is increasing substantially into the second half of 2026, according to investor analysis shared on LinkedIn by Beth Kindig. All four major hyperscalers are expected to see second-half spending between $27 billion and $39 billion higher than their first-half outlays, adding more than $10 billion to quarterly capex bills across the industry. That level of spending creates cascading effects throughout the economy, from semiconductor fabrication to electrical grid capacity to commercial real estate devoted to data center sites.

The Free Cash Flow Problem

What alarmed investors about Meta's earnings wasn't the top-line revenue growth or even the absolute capex number—it was the evaporation of free cash flow. At $784 million for the quarter, Meta generated barely enough cash to fund a single week of its planned infrastructure spending. The company essentially operated at breakeven on a cash basis despite posting record revenue, a dynamic that raises uncomfortable questions about return on invested capital.

Business Insider reported that Meta plans to spend between $125 billion and $145 billion this year, with much of it directed toward data centers and computing infrastructure to power AI ambitions that have yet to produce commensurate revenue. The investment thesis rests on the assumption that generative AI will unlock massive new revenue streams through advertising optimization, content creation tools, and eventually consumer-facing AI products. If that thesis proves wrong or takes longer to materialize than expected, shareholders will have funded the most expensive science experiment in corporate history.

The Robotics Media noted that shares fell approximately 10 percent following the July 29 earnings release despite the record $60.8 billion quarterly revenue figure. The market's message was clear: revenue growth no longer justifies unlimited spending when free cash flow disappears entirely. Investors want to see evidence that the AI infrastructure buildout will eventually translate into profits, not just more ambitious capex targets.

The Supply Chain Windfall

While Meta shareholders absorbed a sharp selloff, the companies supplying AI infrastructure celebrated. Nvidia remains the primary beneficiary, selling the H100 and H200 GPUs that power most large-scale AI training. Custom silicon providers such as Broadcom, which secured a 1 gigawatt AI chip deal with Meta, are seeing unprecedented demand. Data center construction firms, power utilities, and commercial real estate developers focused on hyperscale facilities are booking years of forward revenue based on the spending surge.

Data Center Dynamics observed that a capex range of $130 billion to $145 billion from a single hyperscaler will force suppliers, utilities, and colocation providers to recalibrate their own investment and capacity planning. When one company spends more on infrastructure in a single year than most nations invest in public works, the ripple effects touch every corner of the global supply chain. Blockspace Media emphasized that Meta's revised forecast came as AI infrastructure spending consumed almost all second-quarter operating cash flow, underscoring the capital-intensive nature of the current AI race.

The spending isn't evenly distributed across time. Meta spent $49.1 billion in the first half of 2026, which means the updated guidance implies $81 billion to $96 billion in the second half—a dramatic acceleration that will test whether the supply chain can actually absorb demand at that pace. Semiconductor lead times, data center construction schedules, and power infrastructure upgrades all operate on multi-year timelines, creating potential bottlenecks even when capital is unlimited.

The 2027 Reckoning

Analysts are already projecting over $1 trillion in combined hyperscaler capex for 2027, extending the current trajectory into territory that strains credulity. At some point, the financial returns must justify the investment, or investors will revolt. Meta's stock decline following the July earnings report suggests that patience is already wearing thin, even as the company posts record revenue.

The disconnect between revenue growth and free cash flow generation represents the central tension in Big Tech's AI strategy. Companies are betting that today's infrastructure spending will create tomorrow's monopoly positions in generative AI, but the payoff remains speculative. Meta has yet to articulate how its Llama models or AI-powered advertising tools will generate returns that justify $130 billion to $145 billion in annual capital spending.

Parameter reported that Meta stock dropped 6 percent on speculation about a potential equity offering to fund AI infrastructure, though the company dismissed the reports. The market's reaction to even rumors of dilution demonstrates investor skepticism about whether internally generated cash will suffice to fund the AI buildout without tapping capital markets or slashing other investments.

The AI infrastructure arms race has fundamentally altered how technology companies allocate capital. Shareholders who once received growing dividends and buybacks now watch their ownership stakes fund data centers that may or may not generate competitive returns. Meta's raised capex floor and collapsed free cash flow crystallize the trade-off: Big Tech is choosing AI dominance over near-term profitability, and investors are nervously watching to see if the gamble pays off or becomes a cautionary tale about irrational exuberance in the age of generative AI.