Google PhD Fellowship Research Proposal: 3-Page Guide

Google PhD Fellowship Research Proposal: 3-Page Guide

The Google PhD Fellowship research proposal is a maximum of 3 pages, excluding references, and you cannot send it yourself: students "must be nominated for the fellowship by their university." Universities in the United States and Canada may nominate up to four students. The 2026 cycle ran March 5 to April 30, 2026 and is closed.

That makes two readers, in order. Your department picks its nominees first, then Google reads the file. Google evaluates the proposal next to research impact, academic achievements and leadership potential. Our view: spend at least a third of the three pages on where the work goes next, and name one Google research area. The basis is Google's own sentence: proposals "should include the direction and any plans of where your work is going", and they are judged for relevance "to Google's research areas". A summary of finished dissertation chapters answers neither. The countercase: a nominee close to finishing (Europe and the Middle East accept students "at any stage") has less future to plan. There, the direction section can be the remaining year and the question that follows it, kept short. If you have not separated these documents before, research statement vs SOP vs research proposal explains which one this is: a proposal for work you have not finished.

Every Google rule below comes from its PhD Fellowship program page, checked on October 29, 2026. The page describes the 2026 cycle. Dates for the next cycle were not posted, so recheck it in early 2027.

On this page: who can be nominated · the internal nomination · the three pages · the two essays · constructed example · checklist · questions

Who can be nominated, region by region

The page sets a different cap, stage rule and essay requirement for each region:

For Canada and the United States the award is listed as "Up to 2 year Fellowship" and "US $85K per year towards education costs (tuition and fees), living expenses, travel, and personal equipment (effective from 2025 for new recipients)". Other regions' amounts differ; check your row on the page.

The page lists four eligibility rules for all regions:

• "Students must remain enrolled full-time in the PhD program for the duration of the Fellowship or forfeit the award."
• Google employees and their spouses, children and household members are not eligible.
• "Students that are already supported by a comparable industry award are not eligible. Government or non-profit organization funding is exempt."
• "Past awardees from the PhD Fellowship program are not eligible to apply again."

Separately, "All application materials should be submitted in English."

If you hold or expect other funding, read how external fellowships stack with a PhD stipend before your department nominates you.

Win the internal nomination first

Google's instruction is short: "Please contact your department or graduate school for more information on the internal nomination process." The university then submits "a single, flat (not portfolio) PDF file" for each nominee. For Canada, East Asia and the United States, that file includes:

• Student CV "with links to website and publications (if available)"
• "Research/dissertation proposal (maximum 3 pages, excluding references)"
• "Cover sheet signed by the Department Chair confirming the student passes eligibility requirements"
• "Short (1-page) CV of the student's primary advisor"
• "2-3 letters of recommendation from those familiar with the nominee's work (at least one from the thesis advisor)"
• The two essays, and transcripts

Google also says: "Incomplete proposals will not be considered."

Google does not publish university deadlines, and each university sets its own process. Our recommendation: ask your graduate coordinator in winter how your department picks its nominees, and have a full draft ready before the window opens. In the United States and Canada a university can send at most four files, so the internal pick is a real selection step.

What the three pages must do

This is the whole evaluation statement:

Applications are evaluated on the strength of the research proposal, research impact, student academic achievements, and leadership potential. Research proposals are evaluated for innovative concepts that are relevant to Google's research areas, as well as aspects of robustness and potential impact to the field. Proposals should include the direction and any plans of where your work is going in addition to a comprehensive description of the research you are pursuing.

Split into the parts a reader checks, and mapped to the page (our recommendation):

Google sets no font, spacing or margin rules on the page we checked. It caps length only. Use your field's normal paper format, and keep the references to a separate list after page 3.

Name one research area, in Google's words

The page lists eleven research areas: Computer Architecture; Algorithms and Optimization; Health Research; Human-Computer Interaction; Machine Learning and ML Foundations; Machine Perception; Natural Language Processing; Privacy, Safety, and Security; Quantum Computing; Software Engineering and Programming Languages; and Systems, Networking, and Cloud Computing. It adds that "region-specific research areas will be listed in application forms during the application window" and that "Research should align with Google AI Principles."

Our view: name one area in your first paragraph, using Google's label, and tie your problem to a sentence in that area's call. Do not list three areas to widen the net.

The countercase: interdisciplinary work can honestly sit in two areas. Then pick the area whose call text names your problem most directly, and mention the second once. The same discipline applies in a statement of purpose; see how to make interdisciplinary research coherent.

The two 350-word essays (US, Canada and East Asia)

The page lists both prompts, each with a "350-word limit":

"Describe the desired impact your research will make on the field and society, and why this is important to you. Include any personal, educational and/or professional experiences that have motivated your research interests."

"Describe an example of your leadership experience in which you have positively influenced others, helped resolve disputes, or contributed to group efforts over time. (A leadership role can mean more than just a title...)"

In Latin America the first prompt is the same, and the second asks: "What are your long-term goals for your pathway in computing research, and how would receiving the Google PhD Fellowship help you progress toward those goals in the short-term?" Google says that in these regions "essay responses are evaluated in addition to application materials to determine an overall recommendation."

Our recommendation:

• Impact essay: do not paste the proposal's impact paragraph. This prompt adds "why this is important to you" and the experiences behind it. Use one experience that changed what you work on.
• Leadership essay: the prompt gives three routes: influencing others, resolving a dispute or contributing "over time". Pick one and show it across months, with what changed for other people. Running a reading group or mentoring an undergraduate in your lab can fit; the prompt says a role "can mean more than just a title".

Constructed example: a 3-page outline and one drafted page

Constructed example, written by us, not a real applicant's proposal. The student, the university and every result are invented.

Context (invented): a third-year PhD student in computer science at "Northgate University" in the United States, coursework finished, working on tail latency in replicated storage systems. Area: Systems, Networking, and Cloud Computing.

Outline (our word budgets for single-spaced pages):

1. Page 1, problem and results so far (about 450 words). Why slow outlier requests matter in replicated stores; why fixed request hedging fails under load; the student's first result and where it was published.
2. Page 2, method and why the results hold (about 500 words). The load-aware hedging policy; the testbed; the baselines; the workload where the policy gave no gain, and why.
3. Page 3, where the work is going (about 300 words). Drafted below.

3. Where this work is going (2027–2029)

My first two years established one result. In a replicated key-value store, sending a duplicate request after a fixed delay cuts tail latency at low load but makes it worse once the cluster is busy, because the duplicates become the load. My load-aware policy kept the tail-latency gain at every load we tested on our 64-node testbed, up to 80% utilization.

The open question is whether the policy can tune itself. Today it needs two thresholds set by hand for each workload.

Year 1. Replace the hand-set thresholds with an online controller that sets the hedging delay from queue signals each replica already reports. Milestone: match my tuned policy on the testbed with no per-workload tuning. Risk: if the controller oscillates under bursty load, I will fall back to a small table of thresholds chosen by load level, which still removes manual tuning for the workloads I have.

Year 2. Test the controller on public cluster traces and on a 256-node allocation of our university's shared cluster, and extend it from reads to writes, where duplicates are not free.

Impact. If hedging can tune itself, operators can leave it on during the busy periods when they need it most, instead of turning it off to avoid overload. The result would apply to any replicated service with a tail-latency target, not only key-value stores.

Fit. This work belongs in Google's Systems, Networking, and Cloud Computing area, which "focuses on exploration of future software and hardware for global distributed systems at unprecedented scale."

The drafted page is 258 words, which leaves room for a figure or a milestone table.

Why it works against Google's sentence: the page states a direction and a plan by year ("direction and any plans"), gives a fallback ("robustness", with the failure case on page 2), says who benefits ("potential impact"), and names one area with a quoted line ("relevant to Google's research areas").

Constructed example: the weak version of the same page

Constructed example, written by us, not a real applicant's proposal.

In my remaining years I will continue to improve tail latency in distributed systems. I plan to explore machine learning approaches and evaluate them on larger systems. I hope my research will have a large impact on industry and society.

It has no question, no milestone, no fallback and no area, so it answers none of the parts of Google's evaluation sentence.

Other industry PhD fellowships

On October 29, 2026, the Microsoft Research PhD Fellowship page still showed this notice: "Microsoft Research has paused our call for proposals/nominations for the 2023 calendar year." It showed no newer cycle. We did not check other companies' programs for this page. If you are a senior or first-year graduate student in a computing-heavy field, the federal DOE CSGF essays and Program of Study guide covers a fellowship you apply to yourself.

Checklist before your department's deadline

1. Find your region's row: nomination cap, stage rule and essays.
2. Ask your department how it picks nominees, and when.
3. Check that you hold no comparable industry award.
4. Write page 1 around one problem and your results so far.
5. Give page 2 a robustness test and the case where your method fails.
6. Give page 3 a plan by year, a fallback and one named research area.
7. Keep the proposal to 3 pages; references go after it.
8. Write the two essays for their own prompts, not as proposal excerpts.
9. Brief your advisor and letter writers on the same direction the proposal states.

Questions about the Google PhD Fellowship proposal

Can I apply for the Google PhD Fellowship directly?

No. "All students must be nominated by an eligible university during the application period; direct applications from students are not accepted."

How long is the Google PhD Fellowship research proposal?

A "maximum 3 pages, excluding references".

When does the next Google PhD Fellowship cycle open?

Not posted on October 29, 2026. The 2026 cycle opened March 5, 2026 and closed April 30, 2026. Universities run their own nomination process and submit during that window.

Does the fellowship lead to a job at Google?

No. The FAQ says recipients "do not become employees of Google due to receiving the award", and that the award "does not preclude future eligibility for internships or employment opportunities at Google, nor does it increase the chances of obtaining them."

Can I hold the Google PhD Fellowship with an NSF or other government fellowship?

The page excludes students already supported by "a comparable industry award", and says "Government or non-profit organization funding is exempt." Your university's own rules on combining awards still apply. Biomedical PhD students in the US often pair industry fellowships with an NIH F31; its F31 candidate section asks for a self-assessment with two to four development areas.

Get a free review of your Google PhD Fellowship research proposal

Get a free Quick Review of your Google PhD Fellowship research proposal against GradPilot's research proposal and statement of purpose rubric. There is no Google-specific rubric; this is the closest match. It reads whether you state a specific research problem early, whether every question has a method that can answer it, whether the plan fits the time you have, and whether your earlier work shows you can do it. For a record-to-next-problem read, use the PhD SOP research dossier rubric; for a standalone proposal, the general PhD research proposal rubric.

More PhD research and fellowship guides:

• GRFP, NDSEG and Hertz in one season: the US federal and foundation fellowships, and what transfers between them.
• NSF GRFP research plan guide: the GRFP's own plan, which follows different rules, if you are also applying to the GRFP.
• Negative results in a PhD statement: how to write up the case where your method did not work.
• PhD essays hub: every PhD statement and proposal guide.
• PhD statement of purpose review: for your program applications.

Sources

Retrieved October 29, 2026, except where noted.

• Google Research, Google PhD Fellowship Program (2026 cycle): application status and dates, nomination rule, research areas, review criteria, award amounts by region, eligibility, required materials, essay prompts, nomination caps, employment FAQ.
• Microsoft Research, Microsoft Research PhD Fellowship: the 2023 pause notice still shown on the page.