At the beginning of June 2026, Alphabet announced it had completed an equity financing totaling $84.75 billion—marking the largest single equity issuance in global history and surpassing the previous record of $70 billion set by Brazil’s national oil company in 2010. Even more noteworthy, this financing is not an isolated event. Alphabet simultaneously raised its 2026 capital expenditure (CAPEX) guidance to $180–$190 billion and signaled that spending would rise significantly again in 2027.
Alphabet’s move is just one snapshot of the broader AI infrastructure race among the four major tech giants. By systematically comparing the 2026 CAPEX budgets, funding structures, and strategic logic of Alphabet (Google), Microsoft, Amazon, and Meta, we can better understand why a traditionally "asset-light" tech company like Alphabet is facing nearly $190 billion in capital expenditures—and why it has returned to equity financing tools not used in over two decades.
Alphabet’s Financing Puzzle: Where Does the $84.75 Billion Come From?
To understand the scale of Alphabet’s financing, it’s essential to break down its structure. The financing consists of three components:
Public offering: $34.8 billion. According to regulatory filings, $18 billion comes from Class A common shares and Class C capital shares, while $16.75 billion is from mandatory convertible preferred shares (issued via depositary receipts). Pricing for these offerings was finalized on June 2, with market demand exceeding supply by several multiples—a rare occurrence for a large, mature company like Alphabet.
ATM (At-the-Market) issuance: $40 billion. This is a continuous mechanism with no fixed issuance date. Starting in Q3 2026, Alphabet will sell Class A and Class C shares directly into the public market as needed.
Berkshire Hathaway private placement: $10 billion. This is the most signal-rich component of the financing. Berkshire acquired $5 billion of Class A shares at about $351.81 per share and $5 billion of Class C shares at about $348.20 per share, representing a 6–8% discount from the closing price before the announcement. Notably, Berkshire began building its Alphabet position in Q3 2025 and, by the end of Q1 2026, held about 58 million shares (roughly $17 billion). With this $10 billion injection, Berkshire’s total stake will rise to the $27–$32 billion range.
Why does Alphabet need such massive external financing? The answer lies in the sheer scale of its CAPEX. Even though Alphabet generated about $174 billion in operating cash flow over the past 12 months, its annual capital spending of $180–$190 billion is now too large to cover with internal cash flow alone. According to Capital Futures, before launching its equity financing, Alphabet had already raised more than $85 billion through the bond market in the past year, spanning six currencies: USD, EUR, GBP, JPY, CAD, and CHF—including the rare issuance of a 100-year GBP bond. This is a classic "full-spectrum financing" strategy: the debt side smooths out long-term cost curves, while the equity side locks in immediate large-scale capital, creating a cross-tenor, multi-currency funding safety net.
2026 CAPEX Panorama: Comparing the Four Tech Giants
The capital race for AI infrastructure in 2026 has evolved from phased investment to a systematic strategic reinvestment. According to Goldman Sachs’ updated forecasts in June 2026, the four hyperscale data center operators (Alphabet, Amazon, Microsoft, Meta) will collectively spend $725 billion on CAPEX in 2026—a 77% increase from $410 billion in 2025. S&P Global’s estimates also point to the $700 billion-plus range.
Here are the specific figures and guidance for each company:
| Company | 2026 CAPEX Guidance | Change vs. 2025 (approx.) | Core Spending Focus |
|---|---|---|---|
| Amazon (AWS) | ~$200 billion | +50% | AI data centers, proprietary chips (Trainium/Graviton), logistics, satellite networks |
| Alphabet (Google) | $180–$190 billion | +100% or more | AI data centers, next-gen TPU R&D, Gemini model training |
| Microsoft (Azure) | ~$190 billion (calendar year) | +130% (annualized) | GPU/CPU clusters, long-term infrastructure (15+ years), energy supply |
| Meta | $115–$135 billion (later raised to $125–$145 billion) | +85% | Meta Superintelligence Labs, third-party cloud infrastructure, AI training clusters |
Sources: TrendForce May 2026 report, Goldman Sachs June 2026 update, S&P Global February 2026 analysis. Note: Microsoft’s data varies by reporting method—TrendForce uses $190 billion for the calendar year, while S&P Global records about $140 billion for the fiscal year ending in June. The difference stems from definitions and accounting treatment.
Amazon leads with guidance of $200 billion. According to its February 2026 earnings report, CEO Andy Jassy stated that AWS is the main investment focus, with a substantial portion allocated to the "Rainier" AI infrastructure project built around proprietary Trainium2 chips. AWS has deployed nearly 500,000 Trainium2 chips, aiming for 30% of AI computing tasks to be handled by proprietary chips by the end of 2026.
Microsoft’s $190 billion in spending contains a key structural insight: in Q2 of fiscal 2026 (Q4 2025 calendar), about two-thirds of CAPEX is directed toward short-lifecycle equipment like GPUs and CPUs, while the remaining third goes to infrastructure with a lifespan exceeding 15 years. This reflects Microsoft’s need to balance two asset classes. More importantly, Microsoft’s "commercial remaining performance obligations" (commercial RPO) have surged to about $625 billion. Even after excluding $281 billion related to OpenAI, the remaining $344 billion is more than twice AWS’s total backlog. This means Microsoft has locked in several years of predictable revenue before its CAPEX is deployed.
Meta’s CAPEX range was revised upward during 2026, from an initial $115–$135 billion to $125–$145 billion. Spending is mainly directed toward its Meta Superintelligence Labs and large-scale data centers supporting AI training. Compared to Amazon and Alphabet, Meta’s absolute CAPEX is smaller, but its year-over-year growth rate of about 85% is the most aggressive among the four giants.
The $10 Billion Signal: Why Is Berkshire Hathaway Getting In Now?
Among Alphabet’s financing components, Berkshire Hathaway’s $10 billion private placement stands out. Berkshire has long been known for "staying away from frontier technology," with its tech holdings mostly concentrated in Apple—which it has often characterized as a "consumer electronics company." This substantial increase in Alphabet investment—following its initial position in Q3 2025 and a more than tripling in Q1 2026—signals a fundamental shift in investment logic.
What is Berkshire betting on?
First, Alphabet’s cash flow "moat" remains solid. Google’s core search advertising business provides a stable, high-margin cash foundation. YouTube is a continually growing ad engine, while Google Cloud is emerging as a second growth curve fueled by AI—with Q1 2026 cloud revenue surpassing $20 billion, up 63% year-over-year, and AI solutions revenue growing nearly 800% year-over-year.
Second, while capital spending on AI infrastructure is "burning cash," the assets generated have long-term value. Once built, data centers typically operate for over 15 years. For Berkshire, which is known for long-term holdings, Alphabet’s investment logic is not fundamentally different from its investment in railroads (BNSF)—one builds long-term infrastructure for freight transport, the other builds long-term compute infrastructure for AI.
Third, the risk-reward profile of entering at a discount is attractive. Berkshire secured its Class A and C shares at a 6–8% discount, acting as a "cornerstone investor" within Alphabet’s $84.75 billion financing. This both alleviates short-term concerns about equity dilution and lends credibility to Alphabet’s fundraising efforts.
Compute and Power: The Two Hard Constraints of the CAPEX Race
To understand the allocation of the $725 billion in CAPEX for 2026, we need to examine two core constraints: compute supply and power capacity.
On the compute side, all the giants are accelerating their shift toward proprietary chips (ASICs). Google’s TPU (Tensor Processing Unit) has reached its seventh generation, becoming the most important alternative to NVIDIA GPUs. Amazon’s Trainium series is also widely deployed within AWS, with 1.4 million Trainium2 chips shipped by early 2026. While Microsoft hasn’t disclosed proprietary ASIC deployments of similar scale, about two-thirds of its Q4 2025 CAPEX went to short-lifecycle GPUs/CPUs, with the rest allocated to long-term infrastructure—showing a clear asset mix strategy. The strategic value of proprietary chips is not to immediately replace NVIDIA, but to reduce dependence on a single supplier and achieve better unit compute costs for inference workloads.
On the power side, AI data center electricity consumption is now measured in gigawatts (GW). In Q2 of fiscal 2026 (Q4 2025 calendar), Microsoft added 1 GW of data center capacity in a single quarter. TrendForce projects that global data center installed power capacity will reach about 155 GW in 2026—a 29% year-over-year increase—with AI server power consumption surpassing that of general-purpose servers for the first time. This means electricity supply—and associated cooling systems and high-voltage direct current (HVDC) transmission technology—are becoming key variables in the CAPEX race.
Long-Term Outlook: The Meaning and Risks of $5.3 Trillion
Goldman Sachs’ June 2026 report raised its cumulative CAPEX forecast for the four hyperscale data center operators from $4.5 trillion to $5.3 trillion for the 2025–2030 period. If this spending were treated as the GDP of an economy, its scale would exceed that of Japan, the UK, India, France, and more than 200 other countries—making it the "world’s fourth-largest economy" after the US, China, and Germany.
This forecast has two key analytical dimensions:
First, financing methods must diversify. According to Goldman Sachs, total industry spending (including data centers, power, and compute) could reach $7.6 trillion over the next five years. Purely relying on operating cash flow can no longer cover such capital needs. Alphabet’s record-setting equity financing, along with Amazon, Microsoft, and Meta’s frequent activity in the bond and private credit markets, all point to a trend: hyperscale data center operators are collectively moving away from the traditional "asset-light, self-financing" model toward a "asset-heavy, multi-channel financing" approach.
Second, the return path is not yet complete. As mentioned earlier, Microsoft has locked in $625 billion in performance obligations, but Alphabet, Amazon, and Meta have less visibility into their long-term revenue. S&P Global notes that while all hyperscale operators are ramping up spending, their ability to absorb these investments without significantly impacting credit metrics varies considerably. The current $725 billion CAPEX scale is approaching one of the largest capital reinvestment cycles in tech history, and the realization period for returns may stretch five to ten years.
Conclusion
The 2026 CAPEX race for AI infrastructure is no longer just a battle of technical approaches—it’s a multidimensional contest across compute, capital, and energy. Alphabet has set a new record with $84.75 billion in equity financing, Berkshire Hathaway has cast a $10 billion vote of confidence during a period of transformation, Amazon leads with $200 billion in guidance, Microsoft leverages locked-in future revenue to support massive CAPEX, and Meta pursues the most aggressive growth rate. The combined $725 billion CAPEX of the four giants in 2026 is just the beginning of Goldman Sachs’ forecasted $5.3 trillion cycle for 2025–2030.
For investors, this cycle brings both opportunity and risk: those who find the optimal balance among compute supply, power capacity, and capital efficiency may secure the most advantageous position in the next AI platform era.
