AI Pulse

In partnership with

Trump to Outline AI Priorities Amid Tech Battle with China

Overview

The Trump administration is poised to publish a landmark artificial intelligence blueprint that promises to ease U.S. regulations and supercharge America’s global AI lead in the face of rising competition from China. President Trump will unveil the plan in a major speech titled “Winning the AI Race,” marking a sharp departure from the more restrictive policies of his predecessor.

Key Policy Changes

• Export Promotion: The blueprint calls for exporting U.S. AI technology abroad, shifting away from the Biden‑era “high fence” approach that limited chip exports to potential adversaries.

• State Law Crackdown: It proposes barring federal AI funding from states with tough AI regulations, aiming to create a uniform national standard that fosters growth.

• Open‑Source Emphasis: A new emphasis on open‑weight and open‑source development seeks to democratize AI tools, allowing domestic and international developers greater access.

Industry and Infrastructure Impact

The plan includes data center initiatives led by the Commerce Department, aimed at streamlining permitting and removing barriers to AI infrastructure expansion. It also calls on the FCC to review state regulations that might conflict with its mandate, ensuring a unified federal approach.

Geopolitical Implications

By easing export controls and promoting open innovation, the U.S. aims to outpace China’s AI investments and maintain both economic and military superiority. Critics warn this could lead to technology transfers that bolster adversarial capabilities, but supporters argue that restricting U.S. AI growth only cedes market share to China.

Future Outlook

Following the blueprint’s release, Trump is expected to sign executive orders to codify many of its recommendations, signaling an aggressive push to position American AI firms at the forefront of a multi‑trillion‑dollar industry. Observers will watch closely to see how quickly federal agencies adapt to these sweeping reforms and whether they trigger similar policy shifts in allied nations.

Alibaba’s New Qwen3‑235B‑A22B‑2507 Beats Kimi‑2 and Offers Low‑Compute Version

Introduction

Alibaba’s Qwen Team has just released Qwen3‑235B‑A22B‑2507, an open‑source LLM that outperforms Moonshot’s Kimi‑2 and even Claude Opus 4 on key benchmarks. This update cements Alibaba’s status as a leading provider of permissively licensed, enterprise‑grade AI models.

FP8 Quantization for Efficiency

The standout feature is the FP8 version, which compresses numerical operations to 8‑bit floating point. This reduces memory usage from ~88 GB to ~30 GB and doubles inference speed, making it viable on single‑node GPUs and local hardware. Enterprises can now deploy powerful LLMs without massive infrastructure investments.

Separation of Reasoning and Instruction Models

Alibaba is moving away from its previous hybrid reasoning approach, opting instead to release dedicated instruct and reasoning models. This yields more predictable responses and substantial gains in benchmark scores across MMLU‑Pro, GPQA, AIME25, and code generation.

Performance Benchmarks

• MMLU‑Pro: Jump from 75.2 to 83.0

• GPQA/SuperGPQA: +15–20 points

• AIME25/ARC‑AGI: >2× improvement

• LiveCodeBench: 32.9 → 51.8 for code tasks.

Enterprise‑Ready Features

Released under an Apache 2.0 license, Qwen3 allows full commercial use, local fine‑tuning via LoRA/QLoRA, and on‑premises compliance auditing. Alibaba also bundled Qwen‑Agent, a lightweight tool‑invocation framework for building agentic systems.

Community and Industry Reaction

AI educators like Paul Couvert and NIK (@ns123abc) on X have praised Qwen3’s benchmark domination and FP8 efficiency, calling it a game‑changer for smaller teams and startups. Hugging Face’s Jeff Boudier highlighted the model’s Azure ML one‑click deployment and local use via MLX on Mac.

What’s Next

Alibaba teases a reasoning‑focused variant and hints at 480B MoE models with 1 million token contexts. Qwen3‑235B‑A22B‑2507 not only raises the bar for open models but signals a shift toward enterprise‑class open‑source AI.

You Don’t Need to Be Technical. Just Informed

AI isn’t optional anymore—but coding isn’t required.

The AI Report gives business leaders the edge with daily insights, use cases, and implementation guides across ops, sales, and strategy.

Trusted by professionals at Google, OpenAI, and Microsoft.

👉 Get the newsletter and make smarter AI decisions.

Lead with AI—no coding needed

OpenAI Seeks Additional Capital From Investors in $40 B Round

Background

After announcing a $40 billion financing round in March, OpenAI is now reopening the round on July 28 to secure the remaining $30 billion needed to fulfill its valuation of $300 billion.

Funding Breakdown

• SoftBank: 75% commitment, now up to $22.5 billion

• Syndicate: Remaining $7.5 billion.

Conditions and Contention

SoftBank’s full commitment hinges on restructuring OpenAI’s governance into a public benefit corporation by year‑end. Tensions have arisen over data center deals and the Stargate project, which aims to deliver 10 GW of AI compute at a projected $500 billion cost.

Investor Mix

OpenAI’s backers span Microsoft, Andreessen Horowitz, Sequoia, Thrive, Coatue, Nvidia, and Reid Hoffman, bringing its total capital raised since 2015 to $63.92 billion.

Strategic Importance

The influx of capital underscores the capital intensity of cutting‑edge AI, funding infrastructure, R&D, and talent acquisition at an unprecedented scale.

Future Implications

With SoftBank’s conditional funding and the restructuring push, OpenAI’s path to public markets or further nonprofit engagement hinges on legal approvals in California and Delaware.

AI Arms Race: US and China Weaponize Drones, Code, Biotech

Introduction

A new Fox News report reveals how the U.S. and China are integrating AI into every facet of modern warfare, from drone swarms to gene‑edited soldiers, raising the stakes for a potential Taiwan conflict.

Drone Swarms and Investment

The U.S. Army’s $36 billion AI overhaul will equip each combat division with ~1,000 drones by 2026, shifting away from crewed aircraft to autonomous systems.

Cyber‑AI Fusion for Preemptive Defense

Experts predict that cyber espionage combined with AI could neutralize threats before they materialize, preventing conflicts through preemptive cyber‑AI strikes.

Biotech on the Battlefield

China’s military has reportedly advanced gene‑editing research, potentially creating enhanced soldiers, while the U.S. focuses on AI‑driven trauma care and synthetic blood to save lives.

Ethical and Strategic Concerns

The U.S. insists on human‑in‑the‑loop for lethal decisions, whereas China may not share the same safeguards, raising ethical red flags in AI‑enabled combat.

The Future of Warfare

With projects like Tesla’s Optimus robot under scrutiny for dual‑use potential, tomorrow’s battlefield may be defined more by algorithms and gene sequences than by tanks and missiles.

Humans Triumph Over AI at Annual Math Olympiad, but Machines Are Catching Up

Competition Results

At this year’s International Mathematical Olympiad, five human contestants achieved perfect 42/42 scores, while Google’s Gemini and OpenAI’s reasoning model each scored a gold‑medal 35/42—the first time AI has hit the threshold.

Significance of the Milestone

This breakthrough shows AI is approaching human‑level reasoning, with natural‑language solvers outperforming previous formal‑language approaches.

Community and Expert Reactions

IMO President Gregor Dolinar praised the clarity and precision of AI‑generated proofs, and experts predict AI‑mathematician collaborations could tackle unsolved research problems within a year.

Computational Demands

Google’s model solved all problems within the 4.5‑hour contest window, a massive improvement over last year’s multi‑day compute for a silver medal, demonstrating both software and hardware leaps.

Looking Ahead

As AI continues to mature, the boundary between machine and human intelligence in math is blurring—setting the stage for a new era of collaborative discovery at the frontier of science.