Alphabet raises $85 billion to bet everything on AI infrastructure

Alphabet this week priced the underwritten portion of an equity raise expected to total approximately $85 billion to fund AI infrastructure it describes as supply-constrained against demand from enterprises and consumers. The announcement, dated June 3, 2026, combined a $10 billion private placement to Berkshire Hathaway, a $30 billion underwritten public offering - which was oversubscribed - and a $40 billion at-the-market program expected to begin in the third quarter of 2026. The underwritten portion alone priced at around $35 billion.

The capital raise is not a standalone financial event. It sits at the intersection of every major strategic shift Alphabet has made since launching Gemini 3 in November 2025, and it carries direct implications for the advertising, search, and cloud markets that most of the marketing industry depends on.

The supply problem behind the numbers

According to Alphabet CEO Sundar Pichai, demand for AI solutions from enterprises and consumers is "meaningfully exceeding" available supply. That framing is the core rationale for raising external equity - a mechanism Alphabet has not used at this scale before. The company generated $174 billion in operating cash flow over the trailing twelve months and ended Q1 2026 with $127 billion in cash and marketable securities alongside $81 billion in debt. Since the quarter closed, Alphabet has issued approximately an additional $20 billion in debt across six major currencies, bringing its pro-forma debt balance to just over $100 billion.

Capital expenditure tells its own story. In 2022, Alphabet spent approximately $31 billion. In 2025, that figure reached $91 billion. For 2026, the company is guiding to a range of $180 to $190 billion - roughly six times the 2022 level and double 2025. According to Alphabet CFO Anat Ashkenazi, the company expects a further significant increase in 2027 beyond the 2026 level. The overwhelming majority of the spend is directed at technical infrastructure.

What does that infrastructure actually consist of? According to Alphabet, it spans 10 million kilometers of terrestrial and subsea fiber connecting more than 30 data centers and over 40 cloud regions. The compute portfolio runs from proprietary Tensor Processing Units and Axion CPUs to NVIDIA GPUs. A decade after introducing its first commercial TPU, Alphabet now operates eighth-generation versions - designated TPU 8i and TPU 8t - built with a dual-chip architecture designed specifically for the training and inference demands of agentic workloads. The distinction between the two chips matters technically: the 8t variant handles large-scale training while the 8i targets inference serving, and the dual-chip approach is positioned as the hardware foundation for running complex multi-step agent tasks at production scale.

Serving cost is the other side of the hardware equation. According to Alphabet, Gemini serving costs fell 78% during 2025 through a combination of model optimizations, efficiency improvements, and utilization gains. Since launching Gemini 3, hardware and engineering work reduced the cost of core AI responses by more than 30%. That trajectory matters for the advertising industry because lower serving costs allow AI features to scale to more surfaces and more query types - including the longer, more complex searches where Google has told advertisers that AI tools are expanding ad delivery.

Token volumes as a measure of scale

One concrete metric that illustrates the pace of change: two years ago, Alphabet's models processed 9.7 trillion tokens per month. As of the investor presentation given on June 3, 2026, that figure is 3.2 quadrillion - an increase of more than 300 times across the company's surfaces, which include AI Overviews, AI Mode, the Agent Platform, AI Studio, the Gemini app, Workspace, Antigravity, and others. Model APIs are now processing approximately 19 billion tokens per minute. Over the past twelve months, more than 375 cloud customers have each processed more than one trillion tokens through Alphabet's models.

More than 8.5 million developers are now building applications and experiences using Alphabet models each month. Antigravity, the agentic development platform launched approximately six months ago, has already attracted millions of developers. It has since been expanded into a unified platform covering agent management, an SDK, a CLI, and a dedicated IDE. A standalone desktop application has been released, described by Alphabet as the first step toward an independent, agent-focused surface. Alphabet is also bringing the Antigravity ecosystem to Google Cloud customers for the first time.

What this means for search and advertising

The numbers that matter most for the marketing community sit inside the search and advertising segments. According to Alphabet's Q1 2026 results, Search revenue grew 19% year-over-year to $60.4 billion in the first quarter, Google Services operating margin expanded to 45% from 42% a year earlier, and total consolidated revenue reached $109.9 billion - the eleventh consecutive quarter of double-digit growth.

AI Overviews now have over 2.5 billion monthly active users. AI Mode, launched approximately one year ago, has already surpassed 1 billion monthly users - a trajectory that PPC Land has tracked through 2025 and into 2026 as the defining surface shift for search advertising. According to Alphabet, search queries reached an all-time high in Q1 2026, with AI-powered features driving more usage rather than less. Gemini's understanding of user intent has, according to Alphabet, "significantly expanded" the company's ability to deliver ads on longer, more complex searches.

That framing is significant. AI Mode searches are on average three times as long as traditional searches. Longer queries carry richer intent signals. And the ad formats being built for those surfaces - Conversational Discovery ads, Highlighted Answers, AI-powered Shopping ads - require AI Max for Search or Performance Max with text customization to access them, as confirmed at Google Marketing Live 2026. The commercial logic is clear: as AI surfaces absorb more query volume, the automated campaign formats become the primary entry point to that inventory.

The Gemini app itself reached 900 million monthly users as of May 2026, more than doubling year-over-year from 400 million in May 2025. Personal Intelligence, a feature that connects search and the Gemini app to a user's Google apps to provide personalized suggestions, is now available in more than 190 countries. Gemini Spark, a personal agent that runs continuously in the background within the Gemini app, was announced at Google I/O 2026. It connects with Google tools and is expected to add third-party integrations via the model context protocol.

For advertisers, the introduction of the Universal Cart and the Universal Commerce Protocol - an open-source standard co-developed with retailers - represents a structural addition to the commerce stack. The Universal Cart allows purchases to be initiated from Search, Gemini, YouTube, or Gmail, and the underlying protocol is designed for shopping agents and systems to communicate with each other. AI Max and Performance Max are positioned as the primary vehicles for reaching users within these agentic commerce flows, continuing the pattern that PPC Land covered extensively with the retirement of Dynamic Search Ads in April 2026.

Google Cloud at $20 billion

Google Cloud generated a record $20 billion in Q1 2026 revenue, growing 63% year-over-year while expanding operating margins from 18% to 33%. Operating income more than tripled to $7 billion. The Cloud backlog nearly doubled quarter-on-quarter, reaching $462 billion at the end of Q1 - with Alphabet expecting to recognize just over 50% of that backlog as revenue over the next 24 months.

According to Alphabet, 75% of Cloud customers are using AI products. Enterprise AI solutions became Cloud's primary growth driver for the first time in Q1. Deal velocity more than doubled in the $100 million to $1 billion range compared to the same period a year earlier. Alphabet is now moving beyond hosted cloud infrastructure to deliver TPUs directly into enterprise customers' own data centers, unlocking what it describes as a previously untapped addressable market and allowing the world's most advanced AI labs to run their heaviest training and inference workloads on the same hardware that powers Gemini.

That on-premises TPU deployment is a meaningful technical and commercial expansion. Until this point, access to Alphabet's tensor processing hardware was mediated through the hosted cloud. Making it available inside customer data centers changes the procurement calculus for large AI labs and regulated enterprises that cannot or prefer not to run sensitive workloads on shared infrastructure.

YouTube, subscriptions, and beyond

YouTube's total annual revenue for 2025 surpassed $60 billion across advertising and subscriptions combined. The platform has led U.S. streaming watch time for three consecutive years, with over 200 million hours of living-room viewing daily. At Google I/O 2026, Alphabet announced Ask YouTube, a Gemini-powered conversational search experience built directly into the platform.

Paid subscriptions across Google's consumer products reached approximately 350 million, driven by YouTube and Google One. Q1 2026 was the strongest quarter on record for consumer AI plans. The subscription tiers currently span Google AI Plus at $7.99 per month, Google AI Pro at $19.99, Google AI Ultra 5x at $99.99, and Google AI Ultra 20x at $199.99. PPC Land covered the February 2026 Q4 results when Gemini App had reached 750 million users and initial CapEx guidance for 2026 was set at $175 to $185 billion - a range that has since been revised upward to $180 to $190 billion.

Across the Other Bets portfolio, Waymo is completing more than 500,000 fully autonomous rides per week across 11 U.S. cities, with international expansion planned. Wing has completed more than one million home deliveries in partnership with Walmart and DoorDash. Isomorphic Labs is applying Alphabet's latest models to drug discovery. And quantum computing is progressing through the Willow Early Access Program.

Nano Banana, Alphabet's image generation model, has been used to generate more than 50 billion images to date. The Gemini 3.5 series has entered production - Gemini 3.5 Flash is currently available, and Gemini 3.5 Pro is expected in June 2026 - with improvements focused on agentic coding, long-horizon tasks, and real-world capabilities.

What the $85 billion signals for the industry

The scale of this capital raise, combined with the operational metrics from the investor presentation, draws a clear line for the marketing community. Alphabet is not hedging on AI. It is doubling - in fact, roughly sextupling since 2022 - its bet on compute as the foundational resource of its next phase of growth. That compute underpins every product layer that advertising runs on: Search, AI Mode, AI Overviews, YouTube, Gemini, the agent platforms, and the cloud infrastructure that enterprise advertisers increasingly use to build their own AI-powered marketing systems.

The distribution of that growth, however, remains contested. As PPC Land has documented consistently since 2025, Google Network revenue - covering AdSense, AdMob, and Ad Manager placements on third-party properties - fell 4% in Q1 2026, the sharpest quarterly decline in recent periods. By the end of 2025, 90% of Google's advertising revenue flowed to the company's own properties, a historic concentration. The compute and AI investment Alphabet is now financing at this scale will further accelerate the flows that are reshaping the open web's economics.

For advertisers managing budgets across Google properties, the practical consequences are already visible: automated formats are becoming mandatory for accessing new inventory, query volumes are rising while click-through patterns change, and the commerce, search, and agent surfaces are converging around Alphabet's own infrastructure rather than the distributed publisher ecosystem. The $85 billion is, among other things, the financing mechanism for that convergence to continue.

Timeline

• November 2025 - Alphabet launches Gemini 3; Antigravity agentic development platform launches approximately six months before the June 2026 investor presentation
• February 4, 2026 - Alphabet reports Q4 2025 results: $113.8 billion in quarterly revenue, Gemini App at 750 million monthly users, CapEx guidance for 2026 set at $175-185 billion
• February 11, 2026 - Google introduces Shopping ads for AI Mode, which at that point had more than 75 million daily active users
• April 15, 2026 - Google declares AI Max out of beta and announces Dynamic Search Ads will automatically upgrade to AI Max in September 2026
• April 29, 2026 - Alphabet reports Q1 2026 results: $109.9 billion in revenue, Search +19%, Cloud +63% to $20 billion, Network advertising -4% to $6.97 billion; eleventh consecutive quarter of double-digit growth
• May 2026 - Big Tech Q1 ad revenues collectively surpass $150 billion; Google I/O 2026 showcases AI Mode surpassing 1 billion monthly users; Gemini Spark and Universal Cart announced
• May 20, 2026 - Google Marketing Live 2026: new AI search ad formats announced; AI Max for Search or Performance Max with text customization confirmed as mandatory for access to AI Mode and AI Overviews inventory
• June 1, 2026 - Alphabet files prospectus with the SEC announcing proposed $80 billion equity capital raise; Berkshire Hathaway confirms $10 billion investment in a private placement
• June 3, 2026 - Alphabet delivers investor presentation by Sundar Pichai and Anat Ashkenazi; underwritten offering oversubscribed; total expected proceeds revised to approximately $85 billion; CapEx guidance for 2026 updated to $180-190 billion

Summary

Who: Alphabet Inc., led by CEO Sundar Pichai and CFO Anat Ashkenazi, with Berkshire Hathaway as a named anchor investor committing $10 billion in a private placement.

What: A combined equity raise expected to total approximately $85 billion, comprising a $30 billion underwritten public offering (oversubscribed, with $35 billion priced and allocated), a $40 billion at-the-market program beginning Q3 2026, and the $10 billion Berkshire Hathaway private placement. The proceeds are directed at capital expenditure to scale AI infrastructure and global compute.

When: The SEC filing was dated June 1, 2026. The investor presentation by Pichai and Ashkenazi was delivered on June 3, 2026. The at-the-market program is expected to begin in Q3 2026.

Where: The capital raise is structured across public equity markets and a private placement, with Alphabet's AI infrastructure spanning more than 30 data centers and over 40 cloud regions globally, connected by 10 million kilometers of terrestrial and subsea fiber.

Why: According to Alphabet, demand for AI solutions from enterprises and consumers is materially exceeding available supply. The company's CapEx has been guided to $180-190 billion for 2026 - six times the 2022 level - and is expected to increase further in 2027. Alphabet has framed the equity raise as a proactive move to optimize financial flexibility while maintaining a strong balance sheet, funding investments through a combination of $174 billion in trailing twelve-month operating cash flow, new debt issuances, and the equity offering.