aceWhat the evidence says.What the public feels.
Ranks visible company impact from daily public-attention data, official developer ecosystems, and company-linked sources in the Truvace record.
- 01
Maker of ChatGPT and the GPT model line, with a large consumer and developer surface.
89 visible impact · Dominant observable footprintAccountability risk: ModerateEvidence confidence: High116 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 282,462 · Official GitHub stars: 778,018 · Record sources: 114 · Gain/problem source weight: 88/80
Measurement note: Measured from 282.5K Wikipedia views in 30 days, 778K stars across 82 official public repositories, and 114 independent sources in the Truvace record.
What we don’t know: Private usage, enterprise revenue mix, and internal incident rates are not publicly observable.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 02
Maker of Claude, with growing enterprise and developer-tool distribution.
84 visible impact · Dominant observable footprintAccountability risk: ElevatedEvidence confidence: High34 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 549,282 · Official GitHub stars: 636,678 · Record sources: 32 · Gain/problem source weight: 17/29
Measurement note: Measured from 549.3K Wikipedia views in 30 days, 636.7K stars across 67 official public repositories, and 32 independent sources in the Truvace record.
What we don’t know: Private usage volumes, customer concentration, and revenue composition are not disclosed.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 03
Copilot and Azure distribute AI across workplace, developer, and cloud products.
59 visible impact · Material observable footprintAccountability risk: ElevatedEvidence confidence: High14 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 46,477 · Official GitHub stars: 2,016,870 · Record sources: 12 · Gain/problem source weight: 7/11
Measurement note: Measured from 46.5K Wikipedia views in 30 days, 2M stars across 95 official public repositories, and 12 independent sources in the Truvace record.
What we don’t know: Licensed seats do not reveal active use or the quality of resulting work.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 04
The compute layer behind much of frontier-model training and deployment.
55 visible impact · Material observable footprintAccountability risk: ElevatedEvidence confidence: Moderate6 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 238,584 · Official GitHub stars: 304,621 · Record sources: 4 · Gain/problem source weight: 0/4
Measurement note: Measured from 238.6K Wikipedia views in 30 days, 304.6K stars across 98 official public repositories, and 4 independent sources in the Truvace record.
What we don’t know: Shipped hardware does not reveal allocation priorities or real utilization rates.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 05
Google’s consolidated AI research and product arm, including Gemini and AlphaFold.
55 visible impact · Material observable footprintAccountability risk: ModerateEvidence confidence: High49 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 37,951 · Official GitHub stars: 176,489 · Record sources: 47 · Gain/problem source weight: 35/29
Measurement note: Measured from 38K Wikipedia views in 30 days, 176.5K stars across 97 official public repositories, and 47 independent sources in the Truvace record.
What we don’t know: Bundled distribution makes organic usage difficult to separate from default placement.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 06
Maker of Grok, distributed through X and a rapidly expanding compute footprint.
39 visible impact · Emerging observable footprintAccountability risk: ModerateEvidence confidence: High26 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 14,404 · Official GitHub stars: 118,356 · Record sources: 24 · Gain/problem source weight: 16/18
Measurement note: Measured from 14.4K Wikipedia views in 30 days, 118.4K stars across 9 official public repositories, and 24 independent sources in the Truvace record.
What we don’t know: Usage outside X and independent enterprise adoption remain thinly documented.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 07
An AI-native answer engine operating in the search and publishing ecosystem.
34 visible impact · Emerging observable footprintAccountability risk: ElevatedEvidence confidence: High7 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 91,398 · Official GitHub stars: 11,275 · Record sources: 5 · Gain/problem source weight: 1/5
Measurement note: Measured from 91.4K Wikipedia views in 30 days, 11.3K stars across 31 official public repositories, and 5 independent sources in the Truvace record.
What we don’t know: Query volume, retention, and publisher-licensing coverage are mostly private.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 08
Llama models and assistant features distributed across Meta’s global platforms.
31 visible impact · Emerging observable footprintAccountability risk: ModerateEvidence confidence: High7 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 21,713 · Official GitHub stars: 74,520 · Record sources: 5 · Gain/problem source weight: 5/5
Measurement note: Measured from 21.7K Wikipedia views in 30 days, 74.5K stars across 8 official public repositories, and 5 independent sources in the Truvace record.
What we don’t know: Downstream uses of open weights and engagement quality inside Meta apps remain hard to measure.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 09
A European model lab with open-weight releases and enterprise deployments.
27 visible impact · Emerging observable footprintAccountability risk: ModerateEvidence confidence: Moderate6 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 47,245 · Official GitHub stars: 9,391 · Record sources: 4 · Gain/problem source weight: 4/3
Measurement note: Measured from 47.2K Wikipedia views in 30 days, 9.4K stars across 12 official public repositories, and 4 independent sources in the Truvace record.
What we don’t know: Commercial deployment scale behind private enterprise contracts is not visible.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
- 10
An enterprise-focused model and retrieval provider with limited consumer visibility.
6 visible impact · Emerging observable footprintAccountability risk: LowEvidence confidence: Moderate3 measured signal/sources · updated 2026-09-1330-day Wikipedia views: 21,726 · Official GitHub stars: 1,736 · Record sources: 1 · Gain/problem source weight: 1/0
Measurement note: Measured from 21.7K Wikipedia views in 30 days, 1.7K stars across 25 official public repositories, and 1 independent source in the Truvace record.
What we don’t know: Most deployments sit behind private contracts and are not publicly observable.
Wikimedia pageview signal · Official GitHub organization · Truvace record matches
How this ranking works (and what it is not)
The tracker ranks visible public footprint using adoption signals, public attention, sourced gains, sourced harms, regulatory pressure, and transparency. It is not a ranking of the “best” AI company. Evidence-backed problems do not subtract from footprint: consequence counts in both directions, and accountability risk is shown as its own reading.
| Component | Weight | Signals |
|---|---|---|
| Public attention | 35% | 30-day English Wikipedia pageviews, normalized across the measured set |
| Developer ecosystem | 30% | stars and public repositories in the company’s official GitHub organization |
| Evidence-backed gains | 15% | independent stored sources attached to company-linked gain readings |
| Evidence-backed problems | 15% | independent stored sources attached to company-linked problem readings |
| Evidence breadth | 5% | unique independent sources in the published Truvace record |
Honest limits: public signals are incomplete — private usage, revenue, and safety data may not be visible. Company self-claims are discounted unless independently supported. Every reading carries its source count, last-updated date, and a confidence label. Public-attention and developer signals are fetched server-side and cached for one day; evidence signals are recomputed from the published record on every request.
Recomputed live from the record · Sep 14, 2026, 6:53 AM