Intrepid's co-founder and CEO don't do corporate gloss. Their opening letter in the Integrated Annual Report gets into what 2025 actually required: the hard calls, the strategy reset, and how a nearly 30% growth year still came with real challenges.

Read their story

America Confronts AI's Escalating Risks: In the wake of the Mythos incident, a laissez-faire approach to AI is increasingly seen as politically untenable and strategically unwise. The implications of unregulated AI have led to significant discussions on policy and regulation. Link

Netherlands Approves Tesla's Full Self-Driving Software for Public Roads: The Netherlands Vehicle Authority has granted official approval for Tesla's full self-driving software on public roads, the first such authorization in Europe. The move could influence other European countries to consider similar deployments. Link

Urgent Need for US-China AI Dialogue, Says Nvidia's CEO: Jensen Huang warned that rising US-China trade and security tensions are blocking crucial coordination in AI research. His comments underscore the geopolitical stakes shaping AI policy. Link

AI Early Diagnoses Error Rate Strikes 80%: Research indicates that large language models should not be relied upon for patient-facing diagnostic reasoning due to an alarming 80% error rate in early medical diagnoses. Experts are urging caution around AI's role in frontline healthcare decisions. Link

Hidden Biases in Language Model Training Exposed: A recent study finds that language models can pass hidden biases to other models even after those biases are scrubbed from training data. The results raise concerns about how reliably developers can sanitize AI systems. Link

TSMC Reports 50% Surge in Profit Driven by AI Demand: TSMC is expected to post a fourth straight quarter of record earnings, with net profit up 50% as demand for AI infrastructure accelerates. The chipmaker remains one of the clearest financial beneficiaries of the AI buildout. Link

Microsoft Expands AI Infrastructure with 30,000 Nvidia GPUs: Microsoft plans to rent 30,000 Nvidia GPUs for use at Nscale's 230-megawatt campus in Narvik. The move highlights the scale of compute spending required to stay competitive in AI. Link

Amazon Eyes $50 Billion AI Chip Market, Aiming to Challenge Nvidia: Amazon is making an ambitious push into AI chips that could scale into a $50 billion business. The effort could reshape competitive dynamics in one of the most important layers of the AI stack. Link

Neuro-Symbolic AI Excels in Long-Horizon Reasoning with Reduced Energy Costs: New results suggest neuro-symbolic AI can deliver strong long-horizon reasoning while using less energy. The approach could influence how future AI systems are architected for performance and efficiency. Link

Exploring New Pathways in AI and Technology: Dame Jennifer Dixon outlines how organizations can better position themselves to capture growing opportunities in AI and technology. The piece focuses on strategic engagement across policy and industry shifts. Link

Investing in Precision Medicine to Combat Cancer: The article argues for greater investment in precision medicine technologies that could let AI help program cancer against itself. It highlights AI's growing promise in healthcare and advanced treatment design. Link

Jensen Huang on Nvidia's Stance on AI Winners: Jensen Huang said Nvidia does not try to pick AI winners because it is not the company's role and because history offers a cautionary lesson. The comments shed light on Nvidia's positioning at the center of the AI ecosystem. Link

Bridging the Gap in Healthcare AI Potential: The gap between what consumer healthcare AI could achieve and what it currently delivers remains difficult to close. The piece examines the barriers slowing broader adoption and deployment in healthcare. Link

Meta Introduces Hyperagents for Self-Improving AI: Meta has unveiled a hyperagent framework that lets AI systems rewrite their own logic and scale across tasks without constant human engineering. The approach points toward more autonomous AI for non-coding work. Link

OpenAI Expands Access to GPT-5.4-Cyber: OpenAI has opened GPT-5.4-Cyber to thousands, broadening access to its purpose-built cybersecurity model. The company also signaled that more powerful generalized models are still on the way. Link

Microsoft Launches MAI-Image-2-Efficient: The Fastest AI for Image Generation: Microsoft has launched MAI-Image-2-Efficient, its fastest image-generation model to date. The release positions Microsoft more aggressively against rivals like Gemini on speed. Link

Mark Zuckerberg Introduces Muse Spark, New AI Model from Meta: Meta has announced Muse Spark, a new AI model emerging from Meta Superintelligence Labs. The launch adds another high-profile signal in the race to ship benchmarked frontier systems. Link