Research/Education/GOOGLx/How Alphabet is Financing the AI Build
# Tradfi

How Alphabet is Financing the AI Build

BloFin Academy09/11/2026

For twenty years Alphabet was the kind of company that never needed anyone's money. Advertising threw off more cash than the business could spend, and the balance sheet filled up with the remainder.

That era ended in 2026, because building AI infrastructure at the scale Alphabet has chosen costs more per quarter than the company earns, and in June it went to the equity market for tens of billions of dollars. A business that was asset-light for two decades is now financing itself more like a utility than a search engine.

The details are on the public record, and they matter to anyone holding Alphabet in any form, because new shares are a claim on the same earnings you already own.


What Alphabet actually raised

The June 2026 capital plan shows how Alphabet is financing the AI build: the Google parent company, a multinational technology conglomerate, announced plans to raise roughly $80 billion through several instruments at once, with proceeds earmarked for capital expenditures to scale AI infrastructure and global compute. For investors, traders, and anyone tracking Alphabet stock or tokenized shares, the important point is that this funding mix is centered on public equity offerings and preferred stock, not just debt, which brings dilution, cash flow, and return-on-AI questions directly into the valuation.

Filed under a shelf registration dated June 1, 2026, the package ran through several instruments at once:

  • $16.75 billion of mandatory convertible preferred stock, in two series. Each series carries a 6.25% dividend and was sold as 167,500,000 depositary shares, each representing a 1/20th interest in a preferred share (source: Alphabet prospectus supplement, June 2, 2026).

  • A common stock offering of 25,459,689 Class A shares alongside a Class C tranche, underwritten by a syndicate led by Goldman Sachs, J.P. Morgan and Morgan Stanley.

  • $10 billion in a private placement to Berkshire Hathaway, agreed June 1, 2026 and disclosed in an 8-K dated June 4 (source: Alphabet Form 8-K).

  • An at-the-market program of up to $40 billion in Class A and Class C stock, sold gradually through a syndicate that started at three managers and had grown to sixteen by August 2026 (source: Alphabet prospectus supplement).

  • Multi-tranche senior unsecured notes, including a ten-tranche dollar offering filed in August 2026 with maturities running from the 2020s out to 2126.

The three fixed components come to $66.75 billion before counting the common offering, which puts the total in the region of $80 billion depending on where the common tranche priced. This article breaks down why Alphabet chose this mix over relying more heavily on debt, how the Berkshire placement fits in, what dilution the new shares create, how stock-based compensation and vendor financing interact with the raise, and what all of that means if you are trading or holding Alphabet securities as AI capex stays front-loaded.


Why the spending outgrew operating cash flow

Alphabet still generates enormous operating cash flow, and the constraint is the shape of the spending needed to support AI development rather than the size of the business; the build is capital intensive.

Capital expenditure has roughly tripled in two years, and the trajectory is visible in the filings:

  • Full-year 2024 capital expenditure came to $52.5 billion.

  • Full-year 2025 reached $91.4 billion, an increase of nearly three quarters in twelve months.

  • Q1 2026 alone absorbed $35.7 billion in a single quarter.

  • Q2 2026 added $44.9 billion, bringing the first half of 2026 to $80.6 billion (source: Alphabet Form 10-Q).

Half of 2026 therefore cost roughly what the whole of 2025 did, and the pace kept rising. In the second quarter Alphabet reported a free cash flow deficit for the first time, because investment outran the cash the business produced (source: Tom's Hardware). Moody's viewed the capital raise as credit positive. The mechanics of that, and what it does to margins, are set out in Alphabet AI capex and cash flow.

The spending is also front-loaded in a way software never was, because the infrastructure required includes ai compute, more compute capacity, and broader digital infrastructure. Data centers, custom silicon, grid connections, and energy have to exist before a single customer uses them, and Alphabet's guidance points to a further increase in the second half of 2026 and into 2027, which is expected to rise further and is aimed at future growth. Revenue is projected to reach $701.1 billion by 2029. Financing bridges the gap between when the money goes out and when the compute earns.


Why equity rather than debt

Alphabet could have borrowed the whole amount, since it starts from a strong balance sheet, holds a high investment-grade rating, and its bonds run to long maturities that suit assets with fifteen to twenty-five year lives, but its capital structure still shapes whether that money is best raised as debt or equity.

Leaning on equity and hybrids instead was a deliberate choice about where risk should sit, and the logic is worth following.

AI infrastructure returns are genuinely uncertain in a way software returns were not, especially given the long horizon for development as well as normal project risk. Utilization can ramp slowly, hardware cycles can shorten, and supply chain delays can slow efforts to expand and increase support needs past the point where the economics assumed projects would be earning. Debt is indifferent to all of that: the coupon is due whether the data center is full or empty, and covenants tighten exactly when results disappoint.

Equity carries no coupon and breaches no covenant, and it absorbs a bad outcome by being worth less, which is a cost borne by shareholders rather than a crisis imposed on the company. Mandatory convertible preferred sits between the two, paying a fixed 6.25% dividend and then converting into common stock, which is why rating agencies give it partial equity treatment. That approach also helps support AI development as the company accelerates investment.

The at-the-market program adds flexibility on top of that. Rather than one large block that moves the price, Alphabet can sell into demand as it appears, across many days and many prices. It dilutes more gently, and it lets the company stop if conditions turn, giving the stock program more flexibility if market conditions change.


The Berkshire placement, and what it tells you

The most informative single transaction in the package is also the smallest of the fixed pieces.

Alphabet sold Berkshire Hathaway 14,212,035 Class A shares at approximately $351.81 and 14,359,656 Class C shares at approximately $348.20, for $5 billion of each and $10 billion in total (source: Alphabet Form 8-K).

That combination is an unusual thing to be able to observe. Both share classes were priced in the same negotiated transaction on the same day, which produces a clean read on what the voting right is worth: $3.61 per share, or about 1.04%. Ordinary market quotes give a spread contaminated by differing liquidity in the two lines; this one is a single counterparty pricing both at once. How that premium behaves the rest of the time is covered in GOOGL versus GOOG share classes.

The identity of the buyer carries its own information. Berkshire is long-duration capital that does not trade its positions, so a block of that size lands in hands unlikely to sell into the market. It reduces the overhang that a placement with faster money would create, and its willingness to take the price is a signal other investors and institutions read.

What it is not is a rescue: Alphabet was under no funding pressure, and a company raising capital from strength gets better terms than one raising it under duress.


What the new shares do to your claim

This is the part that matters most for anyone holding the stock, the GOOGLX/USDT Spot token that tracks it, or the GOOGLUSDT Perpetual that prices the expectation of it.

Alphabet had roughly 12,284 million shares outstanding across all classes at the time of the offering. The Berkshire placement added 28,571,691 of them, which is about 0.23% of the total. The common offering and the eventual conversion of the preferred add more, and the at-the-market program adds gradually over its life.

Dilution is real and it is simple arithmetic, though whether it costs you anything is a separate question and it turns on a single comparison: does the capital generate earnings faster than the share count grows? If $80 billion of AI infrastructure produces returns above Alphabet's cost of capital, earnings per share rise despite the larger denominator, with margin expansion as the upside case. If it does not, existing holders own a smaller slice of a business that spent heavily for nothing.

Nobody can settle that in advance, and the honest position is that the answer arrives over years rather than quarters. What you can do is watch the two series against each other: share count on one side, and cloud revenue and operating margin on the other, since proprietary models and cloud services are the most likely channels through which the returns show up. That comparison is the whole investment case, and it is more informative than any single quarter's capex headline. The framework for it is standard fundamental valuation applied to a company mid-transition.

The share count is one side of that comparison, and the other is what traders are paying to hold the expectation right now. The live price, open interest and funding rate for the GOOGLUSDT Perpetual sit on its market page, and a glance there around a dilution headline shows which way leveraged positioning is leaning before the share count catches up.


Stock compensation and insider sales, which are different things

Two other sources of new shares are routinely confused with the capital raise, and separating them clarifies the picture.

Stock-based compensation genuinely increases the share count, and Alphabet pays AI researchers, chip designers and cloud engineers substantially in restricted stock, and satisfying those grants means issuing shares. It is a non-cash expense in the accounts and a real cost to shareholders, and it runs continuously rather than in one event.

Insider selling under Rule 10b5-1 plans works differently. Executives schedule sales months ahead to diversify holdings and cover tax on vesting shares, and those sales move existing shares between owners. They can look dramatic in a year when the stock is near highs, and they raise nothing for the company and fund none of the capex.

The distinction is worth holding onto, because the two are routinely reported together. Capital-raising issuance funds the build, compensation issuance is a cost of running it, and insider sales do neither.


Vendor financing and the circular question

Part of the AI build supports ongoing AI development through financing arrangements that blur the line between customer and investment, and it has drawn scrutiny across the sector.

The pattern runs in a loop. Alphabet builds infrastructure and designs its own chips as part of a broader technology and digital infrastructure stack, takes an equity stake in an AI company, and supports that company's access to long-term compute. The AI company then buys cloud capacity from Google, and revenue arrives that originated partly as Alphabet's own capital.

The Anthropic relationship is the clearest example: an equity stake, a large committed compute arrangement, and a financing structure covering data center capacity at gigawatt scale, with hardware partners taking part of the risk. It also makes the services relationship more explicit, because Alphabet is not just investing but securing multi-year demand for compute capacity that Anthropic consumes through Google. Structures where an investor's capital and a customer's revenue overlap are examined more generally in BloFin's coverage of treasury and PIPE deals.

The critique is straightforward, and it is that revenue your own balance sheet funded is lower quality than revenue from an unrelated customer, because it can be sustained past the point where it makes sense. The defense is that committed demand is exactly what justifies building capacity years ahead of a market. Both readings are true at once, and the test is whether these customers keep buying once they are funding it themselves.

What it does change is how to think about the company. Alphabet now carries project-finance characteristics alongside its software economics, which makes it more sensitive to interest rates, credit spreads, the price of electricity, and broader supply chain risks than it has ever been.


What this means if you trade Alphabet

The financing structure matters for traders because it shapes expected outcomes and future positioning, whether you hold shares directly or take exposure through tokenized Alphabet.

Three things are worth tracking each quarter:

  • Share count each quarter, which tells you how much of the at-the-market program has been used and how quickly the denominator is growing. Alphabet has historically shrunk it through buybacks, which now run against the issuance.

  • Cloud revenue and margin, which is where the return on all this spending shows up if it shows up anywhere.

  • Capex guidance revisions, since an increase versus the previous estimate means the financing need continues and a reduction can signal either discipline or softer demand, and the market will disagree loudly about which.

  • Where you take the exposure, since BloFin lists the GOOGLX/USDT Spot pair alongside the GOOGLUSDT Perpetual, and the two behave differently around a dilution headline; Microsoft spending can also influence how the market reads Alphabet's build, even when the token is diluted exactly as the share is and the contract prices the expectation of it.

A holder of a tokenized share is exposed to exactly the same arithmetic as a holder of the share, because the token tracks the share and the share is what gets diluted. Choosing one wrapper over another leaves the underlying question entirely intact.

Looking to gain exposure to GOOGL? To get started, you'll need to first create a BloFin account, fund your account with cryptocurrency, and navigate to the GOOGLX/USDT Spot trading page or GOOGLUSDT Perpetual page.


Frequently asked questions

How much did Alphabet raise for its AI infrastructure build?

The June 2026 package combined private placement financing, debt, and public equity offerings: $16.75 billion of mandatory convertible preferred stock in two series, a $10 billion private placement to Berkshire Hathaway, an at-the-market program of up to $40 billion in Class A and Class C stock, and an underwritten common stock offering of 25,459,689 Class A shares plus a Class C tranche. Alphabet separately issued multi-tranche senior notes in August 2026. The fixed components alone total $66.75 billion, putting the overall plan in the region of $80 billion, with proceeds meant to expand AI infrastructure and compute capacity.

Why did Alphabet issue equity when it generates so much cash?

Because the spending is front-loaded and larger than the cash the business produces in the same period. Capital expenditure was $91.4 billion in 2025 and reached $80.6 billion in the first half of 2026 alone, for capital-intensive AI development where the required infrastructure came before revenue, and Alphabet reported a free cash flow deficit in the second quarter. Equity and hybrid instruments also suit uncertain projects better than debt, since they carry no coupon and breach no covenant if AI returns arrive later than planned, a mix that can still support future growth despite near-term pressure.

What did the Berkshire Hathaway placement involve?

Alphabet agreed on June 1, 2026 to sell an affiliate of Berkshire Hathaway 14,212,035 Class A shares at approximately $351.81 and 14,359,656 Class C shares at approximately $348.20, for gross proceeds of $10 billion split evenly between the two classes. It was a private placement under Section 4(a)(2), disclosed in a Form 8-K dated June 4, 2026.

How much does the capital raise dilute existing shareholders?

The Berkshire placement added 28,571,691 shares against roughly 12,284 million outstanding across all classes, which is about 0.23%. The common offering, the eventual conversion of the mandatory convertible preferred and the at-the-market program add further over time. Whether that dilution costs shareholders anything depends on whether the capital generates earnings faster than the share count grows and ultimately delivers margin expansion.

Does the Berkshire deal say anything about GOOGL vs GOOG?

It gives an unusually clean reading. Both classes were priced in the same negotiated transaction on the same day, at $351.81 for Class A and $348.20 for Class C, a difference of $3.61 or about 1.04%. Ordinary market quotes reflect differing liquidity between the two lines, whereas this is a single counterparty pricing both at once, which isolates what the voting right was worth in that deal.

Do stock-based compensation and insider sales fund the AI build?

No, and they work differently from each other. Stock-based compensation issues new shares to satisfy employee grants, which increases the share count and is a real cost to shareholders, but it pays staff rather than funding capital projects. Insider sales under Rule 10b5-1 plans move existing shares between owners, raise nothing for the company and fund nothing. Only the capital raise itself finances the build.


Researched and written by the BloFin Academy editorial team with AI-assisted drafting. All facts independently verified. Primary sources include Alphabet's Form 8-K dated June 4, 2026 for the Berkshire Hathaway private placement terms and share counts, Alphabet's prospectus supplements dated June 2, 2026 and August 6, 2026 for the mandatory convertible preferred, common stock and at-the-market offerings, and Alphabet's Form 10-Q for the quarter ended June 30, 2026 together with XBRL data from its 10-K filings for capital expenditure figures, current as of September 2026.

Nothing in this article constitutes financial advice, and nothing in it is a recommendation to trade Alphabet in any form. Capital structures change, announced programs may be used in full or in part, and the returns on AI infrastructure spending are unproven. Verify current share counts, capital expenditure and outstanding offerings against Alphabet's latest filings before relying on any figure here.