Selling to Alphabet

Alphabet has told investors it expects capital expenditures of 195 billion to 205 billion dollars in 2026, a range Chief Financial Officer Anat Ashkenazi raised on the second quarter earnings call because the company is, in her words, in a supply constrained environment. Alphabet ended 2025 with 190,820 employees and annual revenues above 400 billion dollars for the first time.

That number is the reason vendors queue up. It is also the reason most get nowhere. The overwhelming majority of it is Alphabet buying land, power, data centres, networking gear and its own silicon, not buying software from you. At this company the phrase we could build that is not a negotiating tactic, it is the institutional default, and it is usually said by an engineer you will never meet.

How the buying process actually runs, and why it is unlike most Global 500 companies

You cannot register yourself as a supplier. You have to be invited. This is the single most important operational fact about selling to Alphabet, and it is published in their own supplier help centre. New suppliers enrol through SAP Ariba Proposals and Questionnaires, and the registration questionnaire requires the email address of your Google point of contact. The form validates the domain. Accepted examples listed by Alphabet include google.com, waymo.com, wing.com, x.com, capitalg.com, gv.com, verily.com, sidewalklabs.com and calicolabs.com.

Read that as a seller. Most large companies run a supplier portal you can walk into off the street, which is why so many teams treat vendor registration as a go to market motion. At Alphabet the portal is downstream of the relationship. Everything else on this page follows from that.

Two enrolment channels, both with clocks. Ariba enrolment takes roughly four business days once the registration and tax questionnaires are both submitted. The alternative channel, Embark, is quoted at five to eight business days. If you do not respond to Alphabet inside ten business days of the request, the pending enrolment is closed out and your Google contact has to issue a fresh invitation.

No purchase order, no work. Alphabet's own guidance to newly enrolled suppliers is to wait for a purchase order to be issued before beginning any work. Teams that start delivering on the strength of a handshake discover this the slow way.

Security review is a named, published, technical process. The Google Vendor Security Assessment has its own documentation set, including remediation guidelines, penetration testing guidelines and technical requirements. Part of the assessment is run through the external rating firm SecurityScorecard, and Alphabet may request SOC 2 Type II, SOC 3 or ISO 27001 evidence and may run in house security testing on the integration points for engagements it classes as high risk. Budget real engineering time for this, not a questionnaire afternoon.

There is a second front door most sellers ignore. Alphabet is not only a buyer, it is a channel. Google Cloud Marketplace private offers let you sell to Alphabet's customers, and offers transacted through a channel partner may qualify to draw down against a customer's existing Google Cloud commitment. Being on Marketplace gives Google Cloud field teams a reason to care about your existence, which is a different motion from selling Alphabet a licence and often a faster one.

Who owns what, so you know which part of Alphabet your deal belongs to

You will not be selling to these people. Their remit tells you which segment your deal sits in and whose priorities it has to survive two or three levels down.

Alphabet reports in three segments: Google Services, Google Cloud and Other Bets. Google Services is the overwhelming majority of revenue and holds Search, YouTube, Android, Chrome and Play. Google Cloud grew 48 percent to 17.7 billion dollars in the fourth quarter of 2025. Other Bets is a set of separate companies including Waymo, Verily, Wing, X, GV and CapitalG, each with its own leadership, its own procurement and, as the Ariba domain list shows, its own email domain. Selling to Waymo is not selling to Google. Treating Alphabet as one account is the most common structural mistake.

Names and titles taken from Alphabet's 2026 proxy statement and Alphabet's own disclosures, verified 11 September 2026. Senior roles change. Verify at source before using any name.

You cannot sell into this account from the bottom

A deal that justifies the cost of selling here runs to millions a year, and for anything touching infrastructure far higher. Nobody approves that from the middle of an organisation, and at Alphabet the approval has to clear a second bar most companies do not impose: it has to beat the internal build option. An engineering organisation that produces its own accelerators has a very high opinion of what it can ship in two quarters, and that opinion is frequently correct.

This is where most outbound fails. It aims at the level where replies are easiest, which is exactly where neither the budget nor the authority to overrule an internal team sits. A buying committee here includes at least one person whose credibility depends on arguing that the team could do it themselves.

The layer that matters sits below the published names, and here it is unusually hard to map

The people who actually sponsor a purchase are the product area leads, the engineering directors, the platform owners and the programme leads inside a specific product or a specific Other Bet. Almost none of them appear on a leadership page.

Two things make this harder at Alphabet than at a comparable company. First, the organisation is genuinely federated. Waymo, Verily, Wing, Calico and GV are separate companies with separate leadership, and a champion in one gives you close to nothing in another. Second, internal reorganisations around AI have been frequent and public, and titles move faster than any external list can track. The map exists, but it has to be built one person at a time, and it decays.

The board is a published second degree map, and a small one

Alphabet's board has ten members, which is smaller than most Global 500 boards, and Alphabet publishes exactly where each director sits elsewhere. As of the 2026 proxy statement:

• John L. Hennessy, Chair of the Board, is a professor at Stanford and was President of Stanford University until 2016. He is on the board of trustees of the Gordon and Betty Moore Foundation.
• Robin L. Washington is currently President and Chief Operating and Financial Officer of Salesforce, and was previously Chief Financial Officer of Gilead Sciences. She chairs Alphabet's compensation committee.
• Roger W. Ferguson Jr. is Chief Investment Officer of Red Cell Partners and was President and Chief Executive Officer of TIAA until 2021 and Vice Chairman of the Federal Reserve until 2006. He currently sits on the boards of Corning and Klarna, and chairs Alphabet's audit committee.
• L. John Doerr is Chair of Kleiner Perkins and currently sits on the board of DoorDash.
• R. Martin Chávez is Partner and Vice Chairman of Sixth Street, and was formerly Chief Information Officer and then Chief Financial Officer of Goldman Sachs. He is on the board of the Broad Institute.
• K. Ram Shriram is managing partner of Sherpalo Ventures, currently sits on the board of Yubico, and was formerly Vice President of Business Development at Amazon.
• Frances H. Arnold is a professor at Caltech and a Nobel laureate in chemistry, and currently sits on the board of Generate Biomedicines.
• Larry Page and Sergey Brin, co-founders, and Sundar Pichai complete the board.

Read that as a seller rather than as trivia. If anyone in your network is senior at Salesforce, Corning, Klarna, Red Cell, Kleiner Perkins, DoorDash, Sixth Street, Sherpalo, Yubico, Generate Biomedicines, Stanford, the Moore Foundation or the Broad Institute, there is a checkable two hop path to an Alphabet director.

One caveat particular to Alphabet. As of the April 2026 record date, Larry Page held 27.4 percent of total voting power, Sergey Brin held 25.3 percent, and all executive officers and directors as a group held 54.3 percent. The dual class structure means the founders control the company, so board pressure is a weaker lever here than at a widely held firm. The value of the board map is access to operators below, not influence over the founders.

Your other connector routes

The ex-Googler population is enormous and organised, but not by Alphabet. There is no employer run alumni programme of the kind some large banks operate. The largest community, Xoogler.co, is independent and self reports more than 35,000 members. People who left still know who stayed, and unusually often they now run startups your company already sells to.

Acquired company alumni. Alphabet has absorbed a long list of firms, from DoubleClick and YouTube to Looker, Mandiant and Fitbit. Anyone who worked at one before the deal now has colleagues inside Alphabet, and those relationships predate the integration.

Google Cloud partners and integrators. The partner ecosystem is a route into Alphabet itself, because partner teams can see which internal groups have the problem you solve.

Finding which of those paths actually exists is what a relationship intelligence platform is for. Your company probably has hundreds of thousands of relationships and no idea which of them touches the forty or so Alphabet people who could sponsor your deal.

Then you have to get your own connector to act

This step kills more introductions than the external ask does. Suppose the best path runs through one of your investors to a Kleiner Perkins partner to John Doerr. That is about as strong a path as exists, and close to unrepeatable. You are asking someone to spend a piece of a relationship that took years to build, on you. Ask twice in a quarter and you will not be asked again.

The same logic applies at every level, including asking a customer champion for a warm referral to a peer inside Google Cloud. Two of your reps approaching the same connector about the same account in the same month is common and almost always invisible.

Three rules for the introduction itself

Ask for the product area, not for Alphabet. An introduction to whoever owns third party tooling for a named Google Cloud product is actionable. An introduction to Google is not.

Write the paragraph your connector will forward. They will not write your pitch. Give them something they can send without editing. The forwardable blurb is the whole craft.

A warm path starts the clock, it does not skip it. You still need an invitation into Ariba, a Vendor Security Assessment and a purchase order. What the introduction buys you is a real conversation about build versus buy, with someone who has the problem, months earlier than cold outreach would have produced one.

See the warm paths into Alphabet

Everything above describes the problem: a holding company structure where the buyer is almost always Google. What it does not tell you is which of your own relationships already reaches inside. That is a question about your network, not theirs, and it is the one most teams answer from memory.

Boomerang maps the warm paths your company already has into accounts like Alphabet, across your team, your customers, your board and investors, and your partners. It then drafts the ask, routes it through the right connector and tracks it to a booked meeting. Book a 15-minute walkthrough and see it run against your own target accounts.