Who will win the Democratic nomination for President in 2028?

TL;DR

Gavin Newsom leads my individual forecast at 19%, with Pete Buttigieg and Kamala Harris at 15% each (calculation below). Other receives 24% because several credible contenders sit outside the named options (documented early-state activity). Harris’s entry decision is the largest lever because it changes both her own prospects and Newsom’s path into the race (Newsom’s interview).

Context

This forecast uses the client’s October 12, 2026, 11:53 UTC cutoff. Potential candidates are building relationships through the midterm campaign rather than competing in an established presidential field (AP, October 9). The DNC has scheduled its convention for August 7–10, 2028, leaving roughly 22 months for entry decisions and campaign reversals (DNC announcement, March 2). Under your rules, I forecast the formally nominated individual: withdrawal before nomination can change the answer; replacement afterward cannot.

The approved calendar starts with South Carolina on January 22, 2028, followed by Nevada, New Hampshire, New Mexico, Michigan and Virginia (DNC, August 15). I read that sequence as increasing the value of a broad coalition from the outset. Strong New Hampshire support alone is not a complete nomination strategy.

Evidence

Early polling leadership deserves weight, but entry and campaign performance are separate risks. My modern reference class covers all five nomination cycles from 2000 through 2020 without a Democratic incumbent seeking reelection at the comparable stage. I use one selected fall midterm-year national preference poll per cycle, not a harmonized polling series. The observations span October 23, 1998 through October 7, 2018; eventual nominees are recorded in the FEC’s historical election compilations.

The selected leader won three of five cycles. Among leaders who entered, three of four won; Gore’s nonentry accounts for the difference (Gallup’s contemporaneous withdrawal analysis). This is a tiny sample with repeated people, different ballots and different initial leads. It supports taking Harris seriously, not assigning her a mechanical historical success rate.

The freshest original national survey I verified was Focaldata’s October 11 release, fielded October 2–5 among 2,553 adults. Its relevant primary subgroup contains 778 respondents who selected the Democratic primary and rated their participation likelihood at least eight out of ten. Harris received 30% and Newsom 15%, including undecided respondents—not the unverified 40% figure attributed to this release (release; original workbook).

Below is the full identified Focaldata first-choice history for the principal listed contenders. Values are weighted support percentages, not nomination probabilities; subgroup counts are unweighted. January is the earliest wave recovered, not a verified start date for all Focaldata polling. A means adult primary selectors, R means the published registered-voter primary screen, and L means primary likelihood of at least eight out of ten.

These rows are not a clean trend across the entire period. The participation screen, named candidates and availability of an undecided response changed. No equivalent April continuation was verified. I therefore do not interpret the difference between the first and last rows as a measured campaign decline.

Independent recent instruments also put Harris first. Echelon’s September 17–21 survey, published September 23, gave her 24% against Newsom’s 14% among 491 Democrats and leaners; its ballot contained 19 named candidates (original toplines). Harvard CAPS/Harris’s September 26–28 survey gave Harris 38% and Newsom 17%, using a shorter ballot without an explicit undecided response. Its overall sample was 2,200 registered voters, but the Democratic-question sample and exact publication day were not established (original results). These are cross-sectional checks, not interchangeable observations to average mechanically.

Geography supplies counter-evidence. Saint Anselm’s August 17–18 survey put Buttigieg at 25%, AOC at 17% and Harris at 3% in New Hampshire. The reported Democratic-question base was 733 within 1,411 likely voters overall (original tables, page 30). Meanwhile, Harvard’s national racial cross-tabs show a large Harris advantage among Black Democrats and weak Buttigieg support; missing subgroup counts make those directional evidence, not precise South Carolina predictions (Harvard results, page 89).

Backup choices also matter. A Lake Research/FairVote online-panel survey of 800 likely Democratic primary voters, fielded May 6–11 and released May 20, ended with Harris ahead of Newsom 52–48 in its ranked-choice simulation, within the reported sampling margin. It found substantial overlap between Newsom and Buttigieg supporters (original release). I use this as a coalition test, not a simulation of the actual delegate process.

Harris’s support is real; her entry is unresolved. A September 20 Axios canvass found five of eight former advisers expected her to skip the race (Axios). That selected insider sample is not an entry probability. CNN documented genuine indecision alongside continued political calls and campaigning (September 17 reporting). More recent reporting describes Doug Emhoff’s private opposition to another run; that is a reported preference, not a veto or Harris decision (WSJ report).

Newsom’s September 14 pledge not to oppose Harris was explicit and defended under questioning. His rationale was their overlapping supporters and resources, while family agreement remained a separate condition of entry (interview transcript). His South Carolina campaigning and fundraising show purposeful preparation if the opening exists (POLITICO, September 7). I found no verified retraction of the pledge.

Buttigieg has clearer positive entry signals. He said he leaned toward running in an August 3 interview (recorded remarks). Axios reported on October 11 that competitive-seat Democrats welcomed his campaign help; I read this as transferable political value (Axios). His national coalition still needs to extend beyond his strongest early-state constituency.

AOC has a credible progressive route, but the Senate is a substantive competing destination for her ambitions (September 22 interview coverage). Her congressional campaign reported $32.6 million in nominal receipts from January 1, 2025 through June 30, 2026, in filings available at the cutoff. That establishes financial capacity, not a presidential decision (FEC).

Shapiro has a strong electability argument but not yet an equivalent national primary coalition. Quinnipiac’s October 7 release, fielded October 1–5 among 1,159 Pennsylvania likely voters, showed him leading Garrity 59–34 with 63% approval. Those are home-state general-election measurements, not nomination support (Quinnipiac). Whitmer’s May 28 remarks combined no current presidential plans with leaving the door open, rather than a lasting exclusion (Michigan Public). Ossoff’s August 17 interview contained a categorical denial of presidential interest, which deserves a substantial discount despite his polling support (recorded interview).

Other is not an obscure-candidate residual. Beshear’s DGA role and Pritzker’s national operation create substantial organizational paths; AP’s October 9 report attributes $38 million in midterm support to Pritzker’s team (AP). Kelly, Moore, Booker and additional prospects have documented early-state activity (AP, September 28). Their separate chances, plus later emergence, justify a large combined allocation.

I convert this evidence into a conditional scenario model. My judgmental serious-participation assumptions are 42.0% for Harris, 25% for Newsom if Harris participates, and 88% for Newsom if she does not. These are not measured frequencies. Participation includes late or sequential entry; each conditional outcome below also incorporates other candidates’ entry uncertainty and withdrawal before formal nomination.

Every column sums to 100%. Their weights are calculated respectively as 42.0% × 75%, 42.0% × 25%, 58.0% × 88%, and 58.0% × 12%. Multiplying each column by its weight and adding produces the submitted probabilities. No prediction-market signal enters the calculation.

What's non-obvious

A pledge against a head-to-head campaign does not make entry at different times impossible. Harris could become a serious contender, withdraw, and leave an opening for Newsom. Newsom could also start while she remains undecided. Those are possible sequences, not verified plans. Treating their participation events as mutually exclusive would erase them; ignoring his explicit pledge would make the opposite mistake (pledge and rationale).

Money is another misleading shortcut. Ossoff’s reported $77.3 million consists of nominal Senate-campaign receipts covering January 1, 2025 through June 30, 2026—not presidential fundraising through October (FEC). The early-state consultant payments I checked were labeled fundraising consulting, not field organizing (FEC disbursements). This does not rule out a presidential plan. It removes fundraising as a sufficient reason to disregard his explicit denials.

Uncertainties

The next major branch point is the November 3, 2026 midterms, whose outcomes remain future at the cutoff (FEC election calendar). Results can change entry incentives, donor support and electability arguments. Fresh South Carolina and Nevada primary polling, presidential-specific staffing, endorsements and actual candidacy announcements would materially sharpen the estimate. I did not verify a fresh South Carolina preference poll or complete presidential staffing rosters.

Private family and donor reporting is weaker than direct decisions. Future legal or reputational developments are another source of shocks: reporting documents an investigation involving Newsom and his wife, not proof of wrongdoing (Reuters, June 15). Earlier Focaldata coverage and some historical sample sizes remain incomplete, and changes in survey screens limit trend interpretation.

The ranking is sensitive to the assumptions. Holding the conditional table fixed, moving Harris’s participation probability from 30% to 55.0% moves her nomination chance from 10% to 19%, and Newsom’s from 21% to 16%. Those are model sensitivity tests, not confidence intervals. The six-decimal outputs make the arithmetic reproducible; the political uncertainty remains several percentage points.