# NVIDIA and OpenAI Make Power Capacity a Core AI Strategy

*By AI High Signal Digest • August 18, 2026*

NVIDIA and OpenAI’s Ohio AI-factory agreement makes power capacity and long-term infrastructure commitments central to the competition, while local models and agent-system advances reshape deployment economics.

## Top Stories

*Why it matters: AI competition is moving beyond model scores toward the physical capacity, local availability, and operating economics that determine deployment.* [^1][^2]

**NVIDIA and OpenAI are locking in the physical substrate of scale.** At Ohio’s PORTS-Pike campus, NVIDIA and SB Energy are securing land, power, and shell for NVIDIA compute, with OpenAI as tenant; OpenAI will operate a full-stack DSX AI factory. NVIDIA says initial capacity is 4.25 GW, with a possible 3.75 GW extension, while OpenAI’s existing and planned commitments represent about 12 GW through 2030—expandable to roughly 16 GW and $600 billion of NVIDIA compute. The support covers defined lease and power payments plus residual value, not the full site cost, and phases in as facilities come online from 2028 to 2030. This moves NVIDIA upstream from selling chips toward securing the sites and demand that support repeated upgrades. [^1]

**Local models are moving from impressive demos to distribution.** Cline reports that Qwen3.8-27B matched DeepSeek V4-Pro and GPT-5.6 Luna on the Artificial Analysis Intelligence Index—the first local model it says has reached frontier capability. Unsloth says its GGUF build reached 2.7 million Hugging Face downloads and became the platform’s #2 trending model. The combination of a frontier-level third-party score and rapid downloads makes local availability a competitive signal, not just a hardware hobbyist story. [^3][^4]

**Agent evaluation is becoming task- and cost-specific.** After analyzing more than 1.7 million Agent Mode sessions, Agent Arena added cost-per-task/Pareto views and Code, Chat, and Work categories. The leaders diverge: GPT-5.6 Sol leads Code, while Claude Opus 5 variants lead Work and Chat. [^2]

## Research & Innovation

*Why it matters: The strongest technical gains this period come from changing how models are scheduled, trained, and instructed—not simply enlarging them.* [^5]

**Weave treats idle time as a cluster resource.** The scheduler time-multiplexes multiple rollout/training jobs while preserving each job’s on-policy synchronization. In a production-scale evaluation with 328 H20 rollout GPUs, 328 H800 training GPUs, and 200 heterogeneous RL jobs, it reported 1.82–1.99× higher throughput, 1.84× lower provisioning cost than naïve disaggregation, and 100% SLO attainment. Its trace simulations reached decisions for 2,000 jobs in 591 ms versus more than five hours for brute force; the approach is aimed at PPO, GRPO, DAPO, and similar phase-dependent workflows, not fully asynchronous systems. [^5][^6][^7][^8]

**SocialRL shows the value of training for delegation, not generic pleasantness.** The paper’s 4B model was trained across six social environments; on held-out negotiation scenarios it matched or exceeded the GPT-5 family per domain, with 78% of buyer openings anchoring below target versus 3% untrained. Its unified model reached 0.627 average utility across the six environments, above GPT-5.1’s 0.619 and GPT-5.2’s 0.613. These are task-specific results from a v1 preprint, not a claim of general 4B parity. [^9]

**Agent skills appear to stabilize procedures more than supply knowledge.** A paper finds procedural anchoring accounts for 65.7% of cases where skills help, versus 4.5% for explicit knowledge injection; actual-use precision falls from 29.6% to 3.3% as the pool grows from five to 100 skills. In a separate 87-task SkillsBench test, Gemini 3.7 Flash rose from 44.9 to 65.9 and #8 to #2 with curated skill files, at roughly $1.80 per task and 186 seconds—beating the reported cost and latency of Opus 5. [^10][^11][^12][^13]

## Products & Launches

*Why it matters: Agent products are increasingly being embedded inside the software and interfaces people already use.* [^14]

**Gemini 3.7 Flash’s Android computer-use quickstart** uses an ADB screenshot → model planning → normalized-coordinate action loop, without accessibility trees, element IDs, or XPath. The open-source implementation works across native apps, webviews, and dynamic canvas interfaces, positioning the model for UI testing, bug reproduction, and task automation. [^14]

**Cursor’s Origin is live.** The code-hosting platform supports repository hosting, pull requests, review, and deployment alongside Cursor’s agents; GitHub repositories sync bidirectionally and remain the source of truth. Vercel, Buildkite, and Depot integrations are already available. [^15][^16][^17]

## Industry Moves

*Why it matters: Commercial AI is showing both accelerating revenue concentration and unusually large new capital commitments.* [^18][^19]

**Anthropic’s scale is accelerating ahead of its IPO.** Bloomberg reports that its annualized revenue reached $65 billion by the end of July, more than seven times its end-2025 run rate; its latest completed quarter reportedly exceeded $11.5 billion versus $787 million a year earlier, with positive adjusted operating income. [^18]

**Capital is still clustering around infrastructure and interfaces.** The Rundown lists Higgsfield’s $400 million Series B at a $5.4 billion valuation, Groq’s $350 million Series A at $3.5 billion, and Wispr’s $280 million Series B at $2 billion. [^19]

## Quick Takes

*Why it matters: Small product changes show how agent control, pricing, and multi-agent coordination are becoming operational features.* [^20]

- Perplexity Computer now gives each connector an **Allow**, **Always Ask**, or **Deny** setting; recurring runs inherit thread approvals. [^20]
- OpenRouter cut GPT-5.6 Sol pricing by 50%, to as low as $1.25 input/$7.50 output per million tokens on flex. [^21]
- Hermes Desktop’s Bot Mode gives each agent its own role, model, memory, skills, tools, and inter-bot communication. [^22]
- Seedance 2.5 leads Video Edit Arena and now offers native 1080p output with 10-bit color. [^23][^24]

---

### Sources

[^1]: [𝕏 article by @JensenHuang](https://x.com/i/article/2089330332369588224)
[^2]: [𝕏 article by @arena](https://x.com/i/article/2089454679130578944)
[^3]: [𝕏 post by @cline](https://x.com/cline/status/2089425906569977896)
[^4]: [𝕏 post by @UnslothAI](https://x.com/UnslothAI/status/2089541444466090064)
[^5]: [𝕏 post by @ZhihuFrontier](https://x.com/ZhihuFrontier/status/2089288697049211276)
[^6]: [𝕏 post by @ZhihuFrontier](https://x.com/ZhihuFrontier/status/2089288708973621363)
[^7]: [𝕏 post by @ZhihuFrontier](https://x.com/ZhihuFrontier/status/2089288712777551914)
[^8]: [𝕏 post by @ZhihuFrontier](https://x.com/ZhihuFrontier/status/2089288716892385576)
[^9]: [From Passive Delegates to Strategic Negotiators: Reinforcing Social Reasoning in Small Language Models with SocialRL](https://arxiv.org/abs/2608.13787)
[^10]: [Demystifying Agent Skills: Why They Work-Until They Don't](https://arxiv.org/abs/2608.14036)
[^11]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2089507951157670302)
[^12]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2089507952910856559)
[^13]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2089507956660613199)
[^14]: [𝕏 article by @_philschmid](https://x.com/i/article/2089334749718364160)
[^15]: [𝕏 post by @cursor_ai](https://x.com/cursor_ai/status/2089399057659596847)
[^16]: [𝕏 post by @kimmonismus](https://x.com/kimmonismus/status/2089407302600429591)
[^17]: [𝕏 post by @cursor_ai](https://x.com/cursor_ai/status/2089399059488350447)
[^18]: [𝕏 post by @kimmonismus](https://x.com/kimmonismus/status/2089440269762445635)
[^19]: [𝕏 post by @TheRundownAI](https://x.com/TheRundownAI/status/2089443425221288359)
[^20]: [𝕏 post by @AskPerplexity](https://x.com/AskPerplexity/status/2089380410656768053)
[^21]: [𝕏 post by @OpenRouter](https://x.com/OpenRouter/status/2089406144297214339)
[^22]: [𝕏 post by @NousResearch](https://x.com/NousResearch/status/2089429432612147572)
[^23]: [𝕏 post by @arena](https://x.com/arena/status/2089448812159045848)
[^24]: [𝕏 post by @BytePlusGlobal](https://x.com/BytePlusGlobal/status/2088156784250986884)