Less for More: Checking in on New York’s CPF Programs

MIP
AHCP
New York
broadband
The ACLP tracked every New York MIP and AHCP broadband award from first announcement to latest figure. The pattern is consistent: grants revised up, promised reach revised down, and a rising cost to connect each location.
Author

Alex Karras, Michael Santorelli

Published

August 18, 2026

Key Takeaways

  • The ACLP catalogued 34 projects (22 MIP, 12 AHCP) from July 2024 to July 2026.
  • MIP grants have been revised up on 10 projects, down on 7, adding about $38M in upward revisions.
  • Promised reach is falling: MIP −15% locations (~19,000 fewer), AHCP −44% households (~15,000 fewer). 15 projects will serve fewer than first announced.
  • Cost per connection is climbing in both programs: MIP +26% on average (up to +428%), AHCP +91% (up to +301%).
  • Bottom line: about $38M more in MIP and AHCP grants, for roughly 34,000 fewer connections promised than first announced.
NoteFull Report Available

This post summarizes the findings of our Data Mining New York’s MIP & AHCP report, which includes the full per-project observation tables, every source, and a detailed methodology.

Introduction

New York has leveraged over $300 million in federal Capital Projects Fund (CPF) and state funds to underwrite broadband deployment via the Municipal Infrastructure Program (MIP) and the Affordable Housing Connectivity Program (AHCP). The ACLP has been closely tracking these programs, both of which have funded projects that will result in significant overbuilding of existing broadband network infrastructure (see also our earlier Round 1 evaluation and Cortland County analyses).

Over the last few months, the state has begun to alter the funding and scope of many MIP and AHCP projects. None of these changes were announced by the state broadband office, ConnectALL. Instead, the ACLP learned of these significant changes by digging through voluminous meeting packets prepared for the board of directors of Empire State Development, which oversees ConnectALL. Previous ACLP analyses of these programs used data received in response to Freedom of Information requests.

ACLP Efforts to Promote Transparency and Accountability

New York does not report these grants consistently. Rather than a standing record per project, cost and scope figures surface piecemeal across board packets, fact sheets, performance reports, and press releases, each stating a project’s status as of that moment. No single source shows how a grant has changed over time.

To recover that, we gathered every available figure from the public record into one catalog. Three cases are flagged explicitly: awards the board merged (Manhattan and the Bronx), a grant cancelled outright (CNY Open Access), and a grant descoped to meet a federal deadline (Franklin County and DANC, cut from 1,600 planned locations to 2).

Needed Public Funding Keeps Increasing…

Figure 1: Grant cost per project, first announced vs. latest.

10 projects have had their grant revised up by at least 5% since first announced, against 7 revised down. The upward revisions add about $38M to the programs’ combined cost. To cover these increased costs, the state moved unspent funds from the AHCP program to cover the MIP shortfall. This fact was buried in a footnote in the state’s latest CPF Performance Report submitted to the U.S. Treasury in July.

Almost none of this financial engineering buys additional reach, as the next two sections show.

…While Promised Connections Keep Decreasing

Figure 2: Promised reach per project, first announced vs. latest. MIP in locations, AHCP in households.

MIP now plans to pass about 19,000 fewer locations than first announced, a 15% cut. AHCP has fallen further: about 15,000 fewer households, down 44%. Across both programs, 15 projects will reach fewer homes than the day they were announced.

The Result: Higher Costs Per Connection

Dividing each grant by the reach it now promises gives the cost to connect a single location or household.

Figure 3: Cost per connection per project, first announced vs. latest. MIP cost per location, AHCP cost per household.

Aggregated across projects, MIP’s cost to reach each location is up 26%, and as much as +428% for a single project (Oswego). AHCP’s cost per household is up 91%, reaching +301% at the extreme.

Program Cost per connection, first Cost per connection, latest Increase
MIP (per location) $2,113 $2,661 +26%
AHCP (per household) $957 $1,822 +91%

A few projects tell the story

The aggregate trends show up plainly in individual awards.

  • Oswego: locations 10,792 → 2,319 (−79%), while the grant rose $26M → $29.5M (+13%). Far fewer homes, for more money.
  • Schoharie: grant up $30M → $42.4M (+41%), while locations rose only +10% (4,000 → 4,390). Cost far outpacing reach.
  • CNY Open Access: a $26M award cancelled outright, to zero.
  • Franklin (DANC): cut from 1,600 planned locations to 2 to meet a federal deadline.
  • Manhattan & the Bronx (AHCP): households 6,080 → 3,607 (−41%).
  • Buffalo City (AHCP): households 5,033 → 3,353 (−33%).

Why it matters

This analysis of figures otherwise buried in scattered public materials shows a clear trend of scope reduction and budget increases. The data so far from NY’s MIP and AHCP suggest these programs may be falling well short of a prudent and efficient use of limited public funds. That the trend could only be recovered by hand, from fragmented sources, underscores the need for programs like these to have clear, standardized data reporting from the outset.

This lack of transparency makes it difficult to ascertain why the state has made such significant changes to these programs. It is likely due to several intertwined factors.

First, CPF-funded projects must be substantially completed by the end of this year. That the state is radically altering the scope of so many projects might mean that they are unlikely to finish their builds by the deadline.

Second, the Treasury Department allows states to request an extension for CPF projects. It appears that New York received extensions for some AHCP projects earlier this year, which means the deadline for substantial completion of those projects is now June 30, 2027. ConnectALL also requested extensions for certain MIP projects, but the request was still pending as of the end of July 2026. Even so, the state has gone ahead and made substantial changes to many MIP projects. This likely reflects the fact that ISPs have continued investing in and edging out their networks, obviating the need for some of this infrastructure. This dynamic has been evident across the country (LEO locations being removed). Nevertheless, the vast majority of MIP and AHCP projects will still result in wasteful overbuilding of existing infrastructure.

Third, broadband project costs have continued to rise in recent years because of increased demand for materials, tariffs, workforce factors, and other reasons. In addition, ISPs of all ilk are encountering higher than expected ROW, pole, and related access costs. This has changed the economics of many projects, including in New York. MIP and AHCP are unique in that they did not require their grantees to put up a match, so in many cases the full project cost is borne by the state. As a result, the state is now having to bear this burden and might have to pony up additional funds down the line to continue offsetting increased costs.

We tracked 34 projects (22 MIP, 12 AHCP) across every public figure we could source: ConnectALL and ESD board materials, project fact sheets, performance reports, and press releases, from July 2024 through July 2026. Each figure is recorded as one observation (project, source, date) exactly as published. Where a project’s cost or scope changes between sources, that change is the result of interest.

Program totals sum each project’s earliest and latest recorded figure, with cancelled projects falling to zero. “Reach” is service locations for MIP and households for AHCP. Cost per connection divides a project’s grant by its promised reach. Three cases are handled explicitly: awards the board merged are reported as one combined project (Manhattan and the Bronx), a cancelled grant is carried to zero (CNY Open Access), and a grant descoped to meet a federal deadline is shown as a scope collapse but held out of the cost-per-connection figures, where its residual grant over two locations is not a meaningful unit cost (Franklin County and DANC). Full per-project tables and every source are in the full report.