Phase III SBIR/STTR Portfolio Snapshot

Executive Summary

This report compresses the FPDS feed (as accessed through SHELDON) into a Phase III SBIR/STTR roll-up.

Pulled the Arrow dataset with sheldon::arrow_fpds() and filtered to records flagged is_phase_03_sbir_sttr.
Grouped the filtered records by research code to count unique contracts, entities, funding and awarding agencies/offices, while also capturing aggregate obligations and obligation date bounds.
Generated human-friendly research type labels (e.g., “Small Business Innovation Research (SBIR) Phase III”) for each code so the presentation mirrors FPDS terminology.

Phase III SBIR/STTR Summary
Distinct awards, funding offices, and obligations by research code
Research Type	Count Actions	Amount Obligation Total	Count ID Contract Analysis Distinct	Count ID Unique Entity Distinct	Count ID Agency Award Distinct	Count ID Agency Funding Distinct	Count ID Office Award Distinct	Count ID Office Funding Distinct	Date Obligation Min	Date Obligation Max	Name Agency Award Top By Amount Obligation	Name Office Award Top By Amount Obligation	Name Agency Funding Top By Amount Obligation	Name Office Funding Top By Amount Obligation
Small Business Innovation Research (SBIR) Phase III	76,142	$34,405,704,831	17,749	2,810	80	106	990	2,158	Jul 29, 1992	Oct 3, 2025	DEPARTMENT OF THE NAVY	NAVSEA HQ	DEPARTMENT OF THE NAVY	NAVSEA HQ
Small Business Technology Transfer (STTR) Phase III	6,486	$2,595,161,090	1,868	881	62	71	451	677	Jun 6, 2000	Oct 1, 2025	DEPARTMENT OF THE AIR FORCE	BATTLE MNGMNT C3 SSC/BCK	DEPARTMENT OF THE AIR FORCE	F2TS3A SMC RN
Source: FPDS via SHELDON

Who is active: Distinct contract and entity counts immediately expose the depth of participation within each research track.
Where the money flows: Obligations and their date range (minimum/maximum) reveal which research codes remain financially current.
Which offices matter: Top funding and awarding offices highlight the stakeholders most engaged with Phase III performers, providing instant leads for follow-on coordination.

Context: In our 2023 briefing on Phase III lineage detection (see the discussion here), we experimented with topic modeling and targeted REGEX searches across FPDS Phase III descriptions to infer the parent awards.
Learning: Even without modern LLMs, simple pattern-matching could often pinpoint lineage indicators (e.g., references to Phase II contract numbers, solicitation IDs, or technology keywords).
Today’s opportunity: With readily accessible large language models, the same workflow can evolve into an interactive assistant that cross-references narratives, validates matches, and surfaces supporting evidence alongside the quantitative summary shown above.

Data Backbone: FPDS via SHELDON supplies verified contract metadata, enabling structured Phase III roll-ups.
Narrative Layer: Historical topic modeling and REGEX experiments demonstrate that descriptions contain latent lineage signals.
Modern Assist: LLMs can now bridge structured summaries and narrative clues, rapidly confirming lineage hypotheses and spotlighting follow-on opportunities.

Prepared by SHELDON using FPDS contract data.