Alphabet doubles AI capex as spending spree grows

In late 2025, Alphabet spearheaded a significant surge in AI infrastructure investment, raising its 2024 capital expenditure forecast from $85 billion to an ambitious $91–$93 billion. This bold move coincided with a strong Q3 earnings beat—adjusted earnings of $3.10 per share on $102.35 billion revenue, marking its first quarter surpassing $100 billion—driven by robust demand for Google Cloud and a $155 billion backlog. The market responded enthusiastically, with Alphabet’s stock jumping 4% and Wall Street analysts raising price targets, underscoring investor confidence in the company’s strategic pivot from an asset-light advertising model to a more asset-heavy AI infrastructure buildout despite concerns about AI’s impact on core search revenue.

This early surge in AI-related capital expenditures was not isolated to Alphabet but emblematic of a broader industry shift among Big Tech giants. Meta, Microsoft, and Amazon followed suit with staggering capex growth—Meta’s spending up 128%, Google’s 83%, and Microsoft’s 74% year-over-year—allocating a larger share of revenue to AI infrastructure than traditional utilities. Amazon notably expanded its AI chip business by 150% quarter-over-quarter and added nearly 4 GW of new power capacity, fueling a 13% revenue increase and a 10% rise in shares. Investors embraced this asset-heavy approach, viewing these massive upfront investments as foundational utilities of the information age, betting that early and aggressive spending would yield outsized returns akin to the cloud revolution.

By early 2026, Alphabet’s commitment to AI infrastructure had escalated dramatically, with capital expenditures projected to reach up to $190 billion—more than doubling the previous year’s forecast. This aggressive investment strategy coincided with blowout Q1 earnings featuring an 81% profit surge to $62.6 billion and a 22% revenue increase to $109.9 billion, propelled by a 63% jump in Google Cloud revenue to $20 billion. The market rewarded this bold asset-heavy pivot, propelling Alphabet’s market capitalization to an all-time high of $4.4 trillion, doubling within a year. However, this shift also sparked investor scrutiny over capital discipline, as the company balanced rapid AI infrastructure buildout against concerns of potential cloud deceleration.

The so-called 'Magnificent 7' tech giants collectively drove a historic AI spending surge, with hyperscaler capital expenditures projected to hit $725 billion in 2026—a 77% increase over 2025. This unprecedented scale of investment underscores a winner-takes-all market dynamic where only a handful of companies can afford the hundreds of billions annually required to dominate AI infrastructure. Alphabet led this charge with a 63% capex growth rate, signaling strong confidence in AI’s transformative potential, while Microsoft and Amazon also demonstrated robust spending and growth. Meta, however, faced investor pushback for its high capex despite solid revenue growth, highlighting the fine line between aggressive investment and market tolerance in this new asset-heavy era.

Alphabet is spearheading an unprecedented AI compute scale-up, aiming to double its AI compute capacity every six months to achieve a staggering 1000x increase over the next 4-5 years while maintaining cost and energy efficiency. Central to this strategy is the deployment of its seventh-generation Ironwood TPU, which boasts nearly 30 times the power efficiency of its 2018 predecessor, enabling massive growth without proportional increases in operational expenses. This aggressive push also involves reducing reliance on Nvidia chips due to supply constraints, underscoring the critical and costly race to build scalable, high-performance AI infrastructure.

The hyperscale AI infrastructure buildout is accelerating at a breakneck pace, with combined capital expenditures from the 'Magnificent 7' tech giants projected to exceed $725 billion in 2026 and potentially reach $1.5 trillion by the end of 2027. Alphabet alone plans to spend upwards of $205 billion in 2026, reflecting a 63% growth in AI compute investment, while Microsoft, Amazon, and Meta also ramp up spending significantly despite varying growth rates and investor scrutiny. This surge is driven by a winner-takes-all dynamic, as these giants race to dominate AI compute capacity amid soaring demand and component price inflation.

This massive infrastructure expansion is not limited to hyperscalers’ data centers but extends deeply into the semiconductor and energy sectors, marking the largest infrastructure buildout in human history. Semiconductor companies critical to AI compute, such as Micron (+770%), Intel (+483%), AMD (+343%), and Nvidia (+85%), have seen explosive stock gains, while data center and power companies like Vertiv (+256%) and Bloom Energy (+1,647%) also outpace traditional sectors. The ecosystem supporting AI compute is broad and multi-layered, involving key players like Broadcom, Celestica, Samsung, SK Hynix, and Taiwan Semiconductor Manufacturing, all fueling a multi-faceted AI infrastructure scale-up.

Despite the enormous capital outlays causing Alphabet to report negative free cash flow for the first time in its 20+ year public history, the company’s efficient capex deployment and full-stack AI infrastructure—including its strategic TPU systems—position it uniquely to capitalize on supply constraints and outpace competitors. Alphabet’s plan to sell its TPU designs externally by 2027 could create a lucrative new revenue stream, while its capex is strategically distributed across Nvidia GPUs, Broadcom-designed custom silicon, and Micron memory. Meanwhile, hyperscalers have raised approximately $270 billion in long-dated debt to fund this buildout, confident that triple-digit AI sales growth and expanding cloud margins will eventually restore robust free cash flow.

By mid-2026, Jacobs Solutions has emerged as a pivotal engineering and infrastructure supplier driving AI infrastructure expansion beyond traditional tech giants, exemplified by its sole-source EPCM contract with Hut 8 for the 1GW Beacon Point AI data center in Texas. Leveraging advanced digital twin technology to accelerate deployment and mitigate risks, Jacobs not only underscores the rise of engineering firms as key AI enablers but also reflects a broader investment paradigm shift toward 'power-first' greenfield developments addressing critical energy and grid constraints. This strategic positioning has fueled Jacobs’ financial momentum, with a 27% year-over-year backlog surge to nearly $29 billion and AI projects now constituting 11% of adjusted net revenue, prompting multiple raised outlooks and improved profit margins.

The AI infrastructure buildout is rapidly broadening into the physical economy sectors such as energy, manufacturing, construction, and utilities, where capital expenditures are generating a multiplier effect across the broader economy. Analysts highlight that AI-driven spending is fueling record data center construction—projected to reach $323 billion by 2031—and driving demand for power grid upgrades amid bottlenecks in energy supply, permitting, and skilled labor shortages. As Sylvia Jablonsky of Defiance ETFs notes, 'this is not just a software story... it’s an energy supply story,' emphasizing the critical role of energy companies and specialized technology firms in meeting AI’s surging power demands.

This physical economy surge is reshaping investment paradigms, with infrastructure suppliers and engineering firms becoming the new 'picks-and-shovels' beneficiaries of AI’s expansion. Institutional investors like the Abu Dhabi Investment Authority are backing companies such as Innio, whose gas-engine technology powers AI data centers, while firms like Eaton and Corning report explosive order growth and stock gains tied to AI infrastructure contracts. The traditional tech versus industrial divide is blurring as AI fosters collaboration, and investors increasingly recognize the vast, underserved $10 trillion physical markets—spanning chemicals, metal, energy, transportation, and healthcare—where AI and robotics promise scalable automation and latent demand unlocking.

Since late 2025, Big Tech’s capital expenditure strategies have undergone a profound transformation, shifting from asset-light, capital-return-heavy models to heavy reliance on debt financing to fuel AI infrastructure expansion. Alphabet’s return to Europe’s debt markets with a €3 billion bond sale, following Meta’s $30 billion issuance, exemplifies this trend, which aligns with recent tax incentives encouraging sustained investment cycles. While this increased capex pressure may temper capital return programs, it reflects a deliberate pivot to ensure raised capital is actively deployed, echoing historical infrastructure buildouts where debt marked a new investment phase.

By the end of 2025 and into 2026, investors began to sharply differentiate between AI monetizers and manufacturers, recognizing that the market was splintering along lines of profitability and capital intensity. As Dorian Carrell observed, valuing companies like Meta and Google as software plays no longer suffices given their massive investments in GPUs and data centers, which alter their risk profiles. This bifurcation extends to debt financing, where firms like Meta and Amazon maintain net cash positions despite raising significant debt, contrasting with speculative players burning cash on AI infrastructure without clear earnings, a risk highlighted by Julien Lafargue.

Entering 2026, the AI market crystallized into a stark divide between capital-intensive hyperscalers such as Meta and Microsoft, investing hundreds of billions in superclusters and cloud commitments, and rapid-growth software leaders like Palantir and Apple, which are delivering outsized revenue growth with less capex. Tesla’s unique approach of funding a $20 billion AI and robotics expansion through robust energy storage margins further illustrates evolving investment strategies. This environment demands granular fundamental analysis to separate durable investment signals from noise, as evidenced by the mixed performance of high-profile bullish and bearish setups across sectors.

Throughout 2026 and into mid-2027, hyperscaler AI capital expenditures have surged into a supercycle, with Goldman Sachs forecasting $757 billion in 2026 and $920 billion by 2027, while Bank of America projects cloud capex reaching $1.4 to $1.5 trillion. This unprecedented scale is funded through a combination of operating cash flow and favorable long-dated debt issuance totaling approximately $270 billion since early 2026. Despite near-term negative free cash flow—Alphabet alone swung to negative for the first time in over 20 years—analysts remain bullish, citing strong cloud revenue growth, supply constraints, and high operational leverage that underpin long-term ROIC estimates between 25% and 50%. However, market skepticism persists, reflected in subdued shareholder returns and stock volatility, underscoring the critical importance of distinguishing profitable hyperscalers with robust cash flows from riskier neoclouds and speculative players.

The forthcoming decade is poised to witness an unprecedented $5+ trillion AI infrastructure capital expenditure supercycle, marking a decisive shift from software-centric innovation to a compute- and hardware-driven paradigm. As highlighted in early 2026 analyses, this transition is not merely about writing superior code but about mastering the physical buildout—powering, cooling, connecting, and manufacturing the infrastructure that enables intelligence to permeate every industry. This evolution expands AI adoption beyond traditional tech sectors into heavy industries such as manufacturing, energy, and logistics, requiring massive investments in chips, data centers, robotics, and power grids, effectively reindustrializing the global economy.

By 2026, AI infrastructure spending had already entered a historic supercycle, with hyperscalers like Alphabet, Microsoft, Amazon, Meta, and Oracle collectively committing over $700 billion annually, fueling a rapid expansion of data center capacity and semiconductor production. This surge is projected to escalate further, with Goldman Sachs and Bank of America forecasting hyperscaler capex to reach $920 billion in 2027 and potentially $1.5 trillion by the end of that year. Such investments are not only reshaping technology landscapes but also driving extraordinary stock gains for key players like Micron (+770%), Intel (+483%), and Bloom Energy (+1,647%), underscoring the strategic positioning of companies supplying chips, power infrastructure, and cooling solutions.

The AI infrastructure buildout is catalyzing a profound transformation in the physical economy, with AI-driven automation and robotics poised to revolutionize traditionally underinvested sectors such as chemicals, energy, transportation, and healthcare—markets collectively exceeding $10 trillion in annual revenue. This shift is exemplified by autonomous vehicle deployments from Waymo and Hyundai’s robotic ‘Metaplant,’ signaling AI’s transition from information generation to physical action. Government initiatives, including the US Bipartisan Infrastructure Law and EU’s NextGenerationEU fund, are injecting historic capital into upgrading power grids and industrial facilities, addressing critical bottlenecks like power connectivity and permitting delays that currently constrain AI infrastructure deployment.

Despite the massive scale of AI infrastructure spending, market dynamics reveal a complex interplay between rapid technological progress and slower institutional investment adjustments, creating a decade of benchmark arbitrage and cautious optimism about long-term returns. Leading analysts from Morgan Stanley and Goldman Sachs emphasize strong unit economics in GPU rentals and AI hardware, projecting operating cash flows exceeding $900 billion by 2027 and return on invested capital between 25% and 50%. However, sustaining this supercycle depends on realizing substantial AI-driven revenues to justify the trillions in capex, with geopolitical tensions and energy price volatility posing risks. Meanwhile, the U.S. is regaining technological leadership, outpacing China in data center infrastructure for the first time in three decades, as compute-centric investment reshapes global tech and industrial landscapes.