Teahose.
SIGN IN
NEW HERE — WHAT TEAHOSE DOES
We read the entire AI & tech firehose — so you don't have to.
PODPodcastsAll-In, No Priors, Acquired…
NEWNewslettersStratechery, Newcomer…
PAPPapersPhysical AI research
PHProduct Huntdaily launches
VCInvestor ScoutSequoia, a16z, Benchmark…
CLAUDE DISTILLS →
7 reads, 30 sec each — free, 6 AM ET.
+ a live graph of the companies, people & themes underneath.
HOME/THEMES/OPEN-SOURCE REINFORCEMENT LEARNING
// THEME

Open-Source Reinforcement Learning

Organizations releasing open-source frameworks, models, and training recipes for reinforcement learning—spanning RLHF for language models, open RL environments, and open robot learning benchmarks—accelerating community-wide progress.

COMPANIES 34VELOCITY — STABLECAPITAL 28D $36099.4M · 45 DEALS
TOP INVESTORS: nvidia (60) · google (11) · blackstone (8) · general catalyst (7) · meta (7)

CAPITAL FIGURES ARE MEDIA-EXTRACTED ESTIMATES, NOT VERIFIED FILINGS.

Capital surged past $24B in a single week in June
$26.2B · wk of 08-10 ▶2026-06-01 ── 2026-08-31 · WEEKLY
Unclassified mega-rounds dominate at $82B across 73 deals
unknown
$87.6B · 78 DEALS
series a
$15.0B · 23 DEALS
series c
$14.7B · 14 DEALS
seed
$7.8B · 9 DEALS
series d plus
$30.5B · 9 DEALS
Mention momentum
MENTIONS / WEEK · PEAK 287

EXTRACTED FROM 25+ PODCASTS & VC NEWSLETTERS · MEDIA-REPORTED FIGURES, NOT VERIFIED FILINGS

// THE LEAD
▲ STRENGTHENING

Open foundation model releases are weaponizing the ecosystem

NVIDIA's open-sourcing of Cosmos 3—including training frameworks, synthetic data pipelines, and model weights—signals a deliberate strategy to commoditize closed competitors while locking developers into NVIDIA's hardware stack. DeepSeek's continued release of detailed technical reports on MoE architectures and DeepSeek Harness (an open-source agent runtime) extends the same playbook from China. Hugging Face remains the distribution layer for this wave, hosting the models, datasets, and libraries that make open releases actionable. Together, these moves are compressing the moat of closed labs: when foundation model weights are freely available, the competitive advantage shifts entirely to compute, fine-tuning infrastructure, and ecosystem lock-in.

// TRENDS
▲ STRENGTHENINGReinforcement learning is cracking open real-world robotics at scale

Google DeepMind's Gemini Robotics 2 release and NVIDIA's GR00T N1—benchmarked on RTX PRO 6000 and Jetson Thor hardware using Temporal GRPO post-training methods—represent RL moving from simulation into generalist physical AI. Stanford's OpenVLA, UC Berkeley's robot RL algorithms, CMU's robotics research, and Peking University's embodied AI work are all feeding directly into this wave. The AGI countdown being revised to 98% following Gemini Robotics 2 underscores how seriously the research community views this inflection.

Why it matters · Operators building robot learning stacks must now evaluate open RL-trained foundation models (GR00T, Gemini Robotics 2, OpenVLA) as drop-in baselines rather than building from scratch.

▲ STRENGTHENINGStrategic mega-rounds are concentrating capital in open AI infrastructure

NVIDIA's $500B strategic financing via Goldman Sachs and BlackRock—anchored by a GPU securitization mechanism—and the $2B growth round backed by Blackstone, Jane Street, Coatue, and NVIDIA illustrate how open AI infrastructure is attracting financial-infrastructure-scale capital. Weekly deal flow peaked at $24.2B (week of June 15) and $20.3B (July 6), with the 90-day total reaching $66.6B across 48 deals. Capital is not spreading evenly: the 'unknown' stage bucket dominates at $81.9B across 73 deals, reflecting the prevalence of structured/strategic rounds that defy traditional VC categorization.

Why it matters · The financing of AI compute is becoming a capital-markets product, not just a VC activity—investors need to track securitization vehicles and sovereign/institutional co-investors alongside traditional funds.

Axios AI+ · Aug 11Axios Pro Rata · Aug 11Axios Pro Rata · Aug 13Axios Pro Rata · Aug 12
▲ STRENGTHENINGAcademia-industry RL research pipelines are formalizing at scale

The co-authorship of GeoMatch by Maria Attarian spanning University of Toronto, Vector Institute, and Google DeepMind exemplifies a structural trend: frontier RL research is being produced inside hybrid academic-industry teams rather than purely within closed labs. Stanford (OpenVLA), UC Berkeley (Sergey Levine's group), CMU, Peking University, MIT CSAIL, and Shanghai AI Laboratory are all named contributors to the open robot learning and RL ecosystem. Prime Intellect's large-scale autonomous AI research experiments and Nous Research's open-source agentic platform further demonstrate that decentralized, community-driven RL research is now generating publishable, deployable outputs.

Why it matters · Academic labs are no longer just talent pipelines—they are co-producers of open RL infrastructure, giving well-networked research universities disproportionate influence over the next generation of foundation model training recipes.

▲ STRENGTHENINGChinese open-weight models are winning the developer distribution war

DeepSeek's open-source agent runtime (DeepSeek Harness, 130 upvotes on Product Hunt) and Moonshot AI's Kimi K2—a 1-trillion-parameter MoE model achieving state-of-the-art on frontier knowledge, math, and coding benchmarks—are demonstrating that Chinese labs can release competitive open-weight models faster than Western incumbents can close them off. Shanghai AI Laboratory's InternLM and InternVL families add another open-source distribution vector from state-backed Chinese research. The 'did you get an Anthropic or OpenAI offer?' talent benchmark shows Western frontier labs are still winning the talent war, but Chinese open-weight releases are winning the developer mindshare war.

Why it matters · Western AI platform companies face a bifurcated competitive threat: closed Chinese models from the top (matching benchmark performance) and open Chinese weights from below (free to deploy), squeezing the commercial rationale for API rental.

CORROBORATED · 3 SOURCE TYPESproduct_hunt_ai · Aug 1420VC · Aug 13The AI Corner · Aug 12LifeArchitect.ai · Aug 13
// COMPANIES
34 COMPANIES
01
Google
google.com
GROWTH · BLACKSTONE + GOOGLE · SEP 4
475 SIGNALS · LAST SEEN SEP 4, 2026
02
Thinking Machines Lab
thinkingmachines.ai
$1.0B · GROWTH · ACCEL · SEP 4
57 SIGNALS · LAST SEEN SEP 4, 2026
03
Moonshot AI
moonshot.cn
$3.0B · IPO · SEP 4
104 SIGNALS · LAST SEEN SEP 4, 2026
04
xAI
x.ai
GROWTH · SEP 3
138 SIGNALS · LAST SEEN SEP 4, 2026
05
Anthropic
anthropic.com
GROWTH · SEP 3
1282 SIGNALS · LAST SEEN SEP 4, 2026
06
Nvidia
nvidia.com
GROWTH · SEP 3
724 SIGNALS · LAST SEEN SEP 4, 2026
07
Harvey
harvey.ai
SERIES_A · SEQUOIA CAPITAL · AUG 30
75 SIGNALS · LAST SEEN AUG 31, 2026
08
Hugging Face
huggingface.co
139 SIGNALS · LAST SEEN SEP 4, 2026
09
DeepSeek
deepseek.com
$7.4B · GROWTH · AUG 27
106 SIGNALS · LAST SEEN SEP 4, 2026
10
Google DeepMind
deepmind.com
SERIES A · AUG 22
126 SIGNALS · LAST SEEN AUG 29, 2026
11
OpenClaw
STRATEGIC PARTNERSHIP · AUG 11
29 SIGNALS · LAST SEEN SEP 2, 2026
12
Decagon
decagon.ai
30 SIGNALS · LAST SEEN AUG 31, 2026
13
Meta
facebook.com
DEBT/EQUITY · PUBLIC BOND INVESTORS + PUBLIC EQUITY INVESTORS · AUG 11
40 SIGNALS · LAST SEEN AUG 22, 2026
14
Fireworks
fireworks.ai
$1.8B · SERIES C/GROWTH/IPO · AUG 2
61 SIGNALS · LAST SEEN AUG 31, 2026
15
Meta
meta.com
$6M · LOBBYING SPEND Q2 2026 · JUL 24
334 SIGNALS · LAST SEEN SEP 4, 2026
16
Mistral AI
mistral.ai
GROWTH · SAMSUNG · JUL 23
35 SIGNALS · LAST SEEN SEP 3, 2026
17
Nous Research
UNKNOWN · ROBOT VENTURES · JUL 15
4 SIGNALS · LAST SEEN JUL 14, 2026
18
Reflection AI
reflection.ai
$2.5B · SERIES C · NVIDIA + SEQUOIA · JUL 8
15 SIGNALS · LAST SEEN JUL 24, 2026
19
Stanford University
stanford.edu
$64M · SERIES A · NATIONAL GRID PARTNERS + STANFORD UNIVERSITY · MAY 15
92 SIGNALS · LAST SEEN AUG 25, 2026
20
Carnegie Mellon University
cmu.edu
42 SIGNALS · LAST SEEN AUG 26, 2026
21
Ideogram
ideogram.ai
11 SIGNALS · LAST SEEN AUG 22, 2026
22
Black Forest Labs
blackforestlabs.ai
9 SIGNALS · LAST SEEN AUG 22, 2026
23
Hermes
5 SIGNALS · LAST SEEN AUG 22, 2026
24
Stockfish
2 SIGNALS · LAST SEEN AUG 13, 2026
25
Lichess
lichess.org
1 SIGNAL · LAST SEEN AUG 13, 2026
26
Farama Foundation
farama.org
2 SIGNALS · LAST SEEN AUG 12, 2026
27
SUB/WAVE
1 SIGNAL · LAST SEEN JUL 28, 2026
28
Thinkie
4 SIGNALS · LAST SEEN JUL 24, 2026
29
Peking University
pku.edu.cn
17 SIGNALS · LAST SEEN JUL 13, 2026
30
UC Berkeley
berkeley.edu
37 SIGNALS · LAST SEEN JUN 30, 2026
31
MIT CSAIL
csail.mit.edu
8 SIGNALS · LAST SEEN JUN 25, 2026
32
FUTO
1 SIGNAL · LAST SEEN JUN 24, 2026
33
RTK
1 SIGNAL · LAST SEEN JUN 17, 2026
34
Shanghai AI Laboratory
shlab.org.cn
9 SIGNALS · LAST SEEN MAY 28, 2026