Top Stories
Nvidia GTC Closes with $1 Trillion in Orders Through 2027
Nvidia’s GTC 2026 conference wrapped up in San Jose with CEO Jensen Huang announcing purchase orders between the Blackwell and Vera Rubin architectures are expected to reach $1 trillion through 2027—double the previous $500 billion projection.
The week brought a flood of announcements: the Groq 3 Language Processing Unit (from Nvidia’s December acquisition), the Vera Rubin platform shipping later this year with 10x better performance per watt, and Dynamo 1.0, an open-source AI infrastructure operating system. Uber confirmed plans to deploy Nvidia Drive AV software across 28 cities by 2028, while Nissan, BYD, Geely, Isuzu, and Hyundai announced Level 4 autonomous vehicle projects using Nvidia’s platform.
The surprise guest? Disney’s Olaf—a free-roaming robotic character showcasing autonomous AI-powered robotics.
Source: NVIDIA Blog
Hunter Alpha: The Mystery Trillion-Parameter Model Nobody Will Claim
A 1 trillion parameter model called Hunter Alpha appeared on OpenRouter on March 11 with no announcement, no press release, no company claiming ownership. Within days, it was processing billions of tokens—and the AI community still can’t figure out who made it.
The specs align suspiciously well with leaked details about DeepSeek V4: trillion-plus parameters, 1M token context, agent-focused architecture. A “V4 Lite” variant briefly appeared on DeepSeek’s website days before Hunter Alpha launched. But testers note stronger censorship and weaker math performance than typical DeepSeek models.
The same anonymous provider previously released “Pony Alpha,” which turned out to be Zhipu AI’s GLM-5. Whatever Hunter Alpha is, it’s currently free to use on OpenRouter.
Source: Business Today
OpenAI Releases First Open-Weight Language Models Since GPT-2
In a significant strategic shift, OpenAI released gpt-oss-120b and gpt-oss-20b under the Apache 2.0 license—their first open-weight language models since GPT-2 in 2019.
The 120B model achieves near-parity with o4-mini on reasoning benchmarks while running on a single 80GB GPU. The 20B variant matches o3-mini performance and runs on edge devices with just 16GB of memory. Both were trained using reinforcement learning and techniques from o3 and other frontier systems.
This marks a notable change for a company that spent years defending closed-source approaches. The models are available on Hugging Face and through Azure AI Foundry.
Source: OpenAI
Quick Hits
-
Google Workspace: Gemini now generates first drafts in Docs by pulling context from Gmail, Chat, and Drive. Search in Drive gets AI Overviews summarizing your files. Rolling out to AI Ultra and Pro subscribers. Google Blog
-
IBM-Confluent: IBM completed its acquisition of Confluent, bringing real-time data streaming to over 6,500 enterprises including 40% of the Fortune 500. IBM Newsroom
-
xBow Unicorn: Autonomous security startup xBow raised $120M at over $1B valuation. Their AI hacker ranks #1 on HackerOne’s U.S. leaderboard. SecurityWeek
-
90% Don’t Trust AI: A Malwarebytes survey found 90% of respondents worry about AI data collection, with 88% refusing to freely share personal information with chatbots. Malwarebytes
-
State AI Laws: The federal push to preempt state AI regulations faces its first major deadlines—the Commerce Department’s evaluation of “burdensome” state laws and FTC policy guidance were due March 11. Expect litigation over preemption scope. Ropes & Gray
Worth Watching
The Open-Source Race Heats Up: Between OpenAI’s gpt-oss release, Meta’s Scout and Maverick models, and the mystery Hunter Alpha, open-weight models are closing the gap with proprietary systems. Epoch AI research suggests open models now trail state-of-the-art by only about three months on average. The competitive dynamics are shifting faster than many expected.
xAI Rebuilding: Following the SpaceX merger and departures of co-founders Jimmy Ba and Tony Wu, Elon Musk acknowledged xAI “was not built right first time around” and is being rebuilt “from the foundations up.” The $1.25 trillion combined entity is preparing for a SpaceX IPO while restructuring its AI operations.