In 2008, California voted to build 800 miles of high‑speed rail.
Seventeen years and $14.75 billion later — not a single mile is operational.

Voters were promised $33 billion, 96 million riders a year, no operating subsidies, and service by 2020. Today the estimate is $128 billion, the scope has shrunk to 171 miles, and the opening date is the 2030s, at best.

The vote passed by just 665,000 votes. Federal investigators are now asking whether the projections that won it were ever true.

Cost overruns happen on megaprojects. This is not that story.

The firm that wrote the business plan received a billion dollars to manage the project. The consultants doing the work staffed the unit checking the work. The ridership model was debunked by three independent reviews — and used for thirteen more years. The first ten billion dollars was spent without an Inspector General.

Each cost revision came after commitments were secured. Never before.

Here’s how it happened.

THE ESCALATION

$33 Billion → $128 Billion

The green line is what voters were promised. The red is what happened. Every point is an official CHSRA estimate drawn from a published business plan. Hover for details.
280%
Cost increase since 2008
79%
Scope reduction — 800 miles promised, 171 funded
$215M/mi
Spain builds comparable HSR for $11-27M/mi
Note the pattern: each time the scope was cut, the estimate dipped temporarily — then resumed climbing. The 2024 range ($89-128B) is the widest ever reported. Even the Authority has stopped pretending to know what the project will cost.
THE RIDERSHIP MODEL

The Survey That Found What It Was Looking For

A ridership model answers one basic question: who is going to ride this train? To build theirs, CHSRA hired Cambridge Systematics, who surveyed people at airports and rail stations — by definition, people who had already chosen to travel between cities by plane or train. They did not survey drivers, bus riders, or people who stay home.
From this sample, the model predicted that 73 out of every 100 riders would be people who currently drive and would switch to the train instead.
SHARE OF RIDERS PREDICTED TO SWITCH FROM DRIVING
73%vs.11–16%
CAHSR model — California projection
Actual rate — France, Spain, Japan:
mature HSR, dense cities, expensive fuel
Countries with established high-speed rail networks, dense populations, and fuel taxes that make driving expensive get 11–16% of their riders from highways. CHSRA’s model assumed California would achieve 4.5 to 6.6 times that rate.
“A discrepancy suggesting intentional exaggeration.”
Federal Railroad Administration, June 2025
After peer reviewers flagged concerns, Cambridge Systematics made undisclosed changes to the methodology. The model submitted for peer review was not the model used to generate the numbers voters saw. CHSRA continued using it for thirteen years — through three independent critiques calling it unreliable, a 57% cumulative reduction in projected ridership, and a finding by their own peer reviewer in 2022 that the analysis was “not investment-grade.” It was still on file when they signed a $3 billion federal grant agreement in 2024.
THE OVERSIGHT STRUCTURE

The Consultants Who Oversaw Themselves

At its peak, WSP/Parsons Brinckerhoff had 25 consultants for every 1 state employee. WSP held the building lease, ran the IT systems, provided 70% of the workforce, and — critically — staffed the Contract Management Support Unit, the group tasked with overseeing consultant performance. The entity being regulated had become the regulator.
At least seven WSP employees held titled positions within CHSRA — signing official technical memos and presenting to the Board as agency staff. It was, in the words of one state auditor, a “shadow government.”
Then the State Auditor examined 184 tasks WSP had certified as “complete” in project tracking systems.
145of 184
tasks WSP marked as “complete” were not actually finished
79% false completion rate — California State Auditor
The billing pattern tells the same story. Disputed invoices hit $142 million in 2016–2018, dropped briefly to $4.4 million after public exposure and an audit — then surged back to $288 million by 2025. The billing practices responded to scrutiny, not to reform.
“Only 56 contract managers overseeing $5B+ in contracts; only 3 worked full-time.”
California State Auditor, Report 2018-108
THE CHANGE ORDERS

$2.8 Billion Bid. $8.6 Billion Spent.

1,584change orders
turned $2.8B in winning bids into $8.6B in final costs — a 200% increase across three construction packages
The CHSRA Board had delegated unlimited change order authority to the CEO with no dollar cap. The CEO could sub-delegate to staff — who were often themselves consultants. One package grew from $1.37B to $3.69B through staff-level approvals. The Board learned about the cumulative $2 billion overrun only after the money was already spent.
One contractor bid exactly $1,234,567,890 — sequential digits — undercutting the next competitor by $506 million with “cost-saving” proposals that systematically failed. Its package nearly tripled in cost. 594 change orders and a $537 million settlement followed.
The best-performing package had two differences: an independent construction manager and a contractor with 722km of Spanish HSR experience. It still grew 86% — but that was a fraction of the worst package’s 290%.
Internal documents show deliberate structuring. An August 2018 memo from a CHSRA regional director planned splitting a $20.8 million scope into four to five change orders of roughly $5 million each — keeping each below the threshold requiring higher-level review. One $18.6 million change order was approved at 251% of the pre-construction estimate, just below the $20 million oversight trigger. The winning contractor on the worst-performing package, Dragados, had previously paid $7.5 million to settle DOJ allegations of DBE pass-through fraud. Its parent company paid a €57.1 million fine for bid-rigging in Spain.
“CHSRA approved amendments based wholly on the information the contractors reported.”
California State Auditor, Report 2018-108
THE OVERSIGHT GAP

Fifteen Years Before Anyone Looked at the Books

$10B+spent between 2008 and 2023 with no Inspector General, no independent oversight body, and no mechanism to catch the disputed invoices, false completions, and funding gaps accumulating across the program
When an IG was finally created in September 2023, it immediately found a $6.5–7 billion funding gap. The gap was not new. It had merely been newly acknowledged.
Meanwhile, the Legislature was moving in the opposite direction. In 2016, AB 1889 was gut-and-amended in the last hour of the legislative session — no video, no public notice, no registered opposition — to redefine the voter-approved “suitable and ready” standard for bond spending. In January 2026, AB 1608 was proposed to allow the new IG to withhold records describing “weaknesses” — including fraud-detection controls.
One day after AB 1608’s language was released, the CHSRA Board approved a $537 million settlement — the largest change order in project history — in closed session. The administration called the timing coincidental.
Then they signed a $3 billion federal grant agreement.
THE STRONGEST CASE — FRAUD SCORE: 5/5

The FSP Agreement

On September 30, 2024, CHSRA signed a Full Funding Grant Agreement obligating $3.07 billion in federal funds. To get it, they committed to completing Merced-Bakersfield by 2033, represented financial capacity to deliver, and submitted ridership projections based on the same model three independent reviews had called unreliable.
They signed despite a known $5-10 billion funding gap. Despite an electrification budget at 22% of what the same state had just paid for comparable work. Despite a rolling stock procurement deadline they could not possibly meet.
“CHSRA could not have made that commitment in good faith.”
Federal Railroad Administration, June 4, 2025
The FRA’s 315-page compliance review — based on 80,000 pages of project documents — found nine specific violations. By July 2025, the FRA terminated approximately $4.2 billion in remaining grants. By February 2026, Congress had rescinded another $929 million.
This is the prosecutorial centerpiece. Under the False Claims Act, organizational liability applies with a “reckless disregard” standard — no need to prove any individual intended to deceive. Continuing to use a model that UC Berkeley (2010), the Peer Review Group (2010), and CHSRA’s own reviewer (2022) all called unreliable meets that standard. Treble damages apply. On $6.9 billion in federal obligations, potential exposure exceeds $20 billion.
ACTIVE AS OF FEBRUARY 2026

Three Federal Investigations

FBI
Congressional referral · March 2025
Rep. Kevin Kiley referred CHSRA to the FBI following the FRA compliance review.
House Oversight
Investigation opened · August 2025
House Committee on Oversight launched a formal investigation into federal fund usage.
Senate Commerce
Report issued · February 2026
Senate Commerce Committee published findings on misrepresentations in grant applications.
ORIGINAL ANALYSIS

Connections Not Previously Documented

Most of this work is synthesis — bringing together scattered public records that have never been systematically cross-referenced. But in the process, we identified the following connections that, to our knowledge, have not appeared in published analysis. We document what we found and what we didn’t.

PROCUREMENT & CHANGE ORDERS
1.

CEO bonus tied to low bid, 65 days before procurement rule change

Resolution #1228 contractually linked CEO Jeff Morales’s bonus to “getting a bid at or below estimates.” Sixty-five days later, the 4th RFP Addendum reversed the Board-approved two-step selection process to favor lowest price over technical competence. The Board was not notified of the change.

2.

Internal memo documents potential criminal change order splitting

A CHSRA regional director’s August 2018 memo explicitly planned splitting a $20.8 million scope into four to five change orders of roughly $5 million each — keeping each below the threshold requiring higher-level review. California Public Contract Code Section 20116 makes splitting work orders to evade oversight a criminal violation.

3.

Threshold-adjacent change order at 251% of independent estimate

An $18.6 million change order was approved at 251% of the pre-construction manager’s independent cost estimate, positioned just below the $20 million Board Operations Committee review trigger. The pricing was not questioned.

4.

Independent PM oversight eliminated during procurement changes

TY Lin International’s independent program management oversight role was terminated in the same period as the CP1 procurement rule change, removing the one entity positioned to flag irregularities in the process.

5.

CP4 as natural experiment isolating governance failure

CP4 had an independent construction manager and a contractor with 722km of Spanish HSR experience. Same state, same project, same era — but 86% growth vs. 290% for the worst package. The comparison isolates procurement and oversight as the variable, not “inherent” US cost factors.

ORGANIZATIONAL CAPTURE
6.

Seven named WSP employees held titled positions within CHSRA

Tony Daniels, Gary Griggs, Kristina Assouri, Hans Van Winkle, Brent Felker, Jim Van Epps, and Daniel Horgan held official agency titles while employed by WSP. Daniels signed CHSRA technical memos. Griggs presented to the Board as staff. Horgan, as Deputy COO, co-authored budget memos with spending authority.

7.

Untested legal precedent for challenging WSP embedding

Professional Engineers in California Government v. Department of Transportation (1997) established that state agencies cannot delegate inherently governmental functions to private contractors. This framework has never been applied to the CHSRA–WSP relationship, despite WSP staffing the unit tasked with overseeing WSP.

FEDERAL GRANT FRAUD
8.

13-year knowledge chain establishes reckless disregard

UC Berkeley (2010), the Peer Review Group (2010), and CHSRA’s own reviewer (2022) all called the ridership model unreliable. CHSRA submitted projections based on the same model in the 2024 FSP Agreement. Under the False Claims Act, this chain establishes “reckless disregard” without needing to prove any individual intended to deceive.

9.

Electrification budget at 22% of the same state’s actual costs

CHSRA budgets $11.8 million per mile for electrification. Caltrain — same state, same regulatory environment, same era — spent $53.3 million per mile. The budget is 22% of comparable actual costs. The FRA flagged it as unrealistic in its compliance review.

VOTER DECEPTION
10.

Post-peer-review model changes went undisclosed

After independent peer reviewers flagged concerns with the ridership model, Cambridge Systematics made undisclosed changes to the methodology. The version submitted for peer review was not the version used to generate the numbers voters saw. No one outside the firm reviewed the final version.

LEGISLATIVE PATTERN
11.

Progressive records shielding across a decade

AB 1889 (2016) was gut-and-amended in the final hour of the legislative session — no video, no public notice, no registered opposition — to redefine standards for bond spending. AB 1608 (2026) would let the new IG withhold records describing “weaknesses,” including fraud-detection controls. The pattern: each legislative intervention weakened oversight, with procedural moves designed to minimize public scrutiny.

METHODOLOGY

We Show Our Work

This investigation documents 77 verified findings across six threads, each tested against the five-element legal standard for fraud. 21 contradictions are flagged and tracked. 205 entities are mapped in a typed knowledge graph that traces the relationships between people, organizations, contracts, and money flows.
Every claim cites a source, tiered by reliability: primary documents, government reports, investigative journalism, and think-tank analysis. Where the evidence is ambiguous, we say so. Where counter-arguments have merit, we present them. All datasets and source code are freely available.
Beyond bringing scattered evidence together, this investigation identified novel connections and legal frameworks — from a CEO bonus contractually tied to a low bid 65 days before procurement rules were changed, to an internal memo that may document a criminal violation of California’s public contract code.