March 20, 2026 – AI Daily Recap

The AI world never sleeps, and this week has been especially relentless. Enterprise spending is shifting at a pace nobody predicted, new models are dropping with capabilities that were theoretical months ago, and a billion-dollar startup just launched to challenge the entire LLM paradigm. Here’s what matters most.

Anthropic Captures 73% of First-Time Enterprise AI Spending

Anthropic is now capturing over 73% of all spending among companies buying AI tools for the first time, according to customer data from Ramp. Just 10 weeks ago, the split with OpenAI was 50/50 — and it was 60/40 in OpenAI’s favor as recently as early December. Meanwhile, Bloomberg reports that Anthropic is nearing a $20 billion annualized revenue run rate. The speed of this reversal signals a genuine inflection point in how enterprises are choosing their AI stack.

OpenAI Launches GPT-5.4 With 1M Token Context and Computer Use

OpenAI introduced GPT-5.4 on March 5 with a 1-million-token context window and native computer use capabilities — the model can read screens, control a mouse and keyboard, and automate workflows across applications. According to TechCrunch, the model produces 33% fewer hallucinations than its predecessor and scored 83.0% on the GDPVal benchmark, placing it at or above human-expert level on economically valuable tasks. Later in the month, OpenAI followed up with GPT-5.4 mini and nano for high-volume workloads.

Google Drops Gemini 3.1 Flash-Lite at $0.25 Per Million Tokens

Google launched Gemini 3.1 Flash-Lite on March 3, delivering 2.5x faster time-to-first-token and 45% faster output generation compared to earlier Gemini versions — all at just $0.25 per million input tokens. As VentureBeat notes, it matches Gemini 2.5 Flash quality while costing a fraction of the price. The model also adds expanded thinking support with configurable reasoning depth, making it a serious option for cost-sensitive, high-volume deployments.

Yann LeCun’s AMI Labs Raises $1.03 Billion for World Models

Turing Award winner Yann LeCun’s new startup, Advanced Machine Intelligence (AMI) Labs, closed a $1.03 billion seed round — the largest ever for a European company. Backed by Nvidia and Bezos Expeditions at a $3.5 billion pre-money valuation, AMI Labs is pursuing “world models” — AI that learns by understanding physical reality rather than just language. As The Next Web puts it, this is a fundamentally different bet on the future of AI, and the funding signals serious institutional confidence in alternative approaches to LLMs.

Meta Unveils Four Generations of Custom AI Chips

Meta announced the MTIA 300, 400, 450, and 500 — four new generations of custom AI chips designed to reduce dependence on Nvidia. According to CNBC, the MTIA 300 is already in production, with the remaining chips rolling out roughly every six months through 2027. Built on RISC-V architecture and manufactured by TSMC in partnership with Broadcom, the lineup doubles HBM bandwidth between generations and is aimed at powering everything from content ranking to generative AI inference at scale.

Microsoft Weighs Legal Action Over Amazon-OpenAI Cloud Deal

Microsoft is reportedly considering legal action over a $50 billion arrangement that makes AWS the exclusive third-party cloud provider for OpenAI Frontier, the company’s new enterprise AI agent platform. The dispute hinges on whether this violates Microsoft’s exclusive Azure hosting agreement with OpenAI. According to PYMNTS, the three companies are still in talks to resolve the matter without litigation, but a lawsuit could disrupt OpenAI’s planned IPO.

The Big Picture

Enterprise buyers are switching AI providers faster than anyone predicted. New model releases are arriving at a pace that makes quarterly planning feel outdated. And real money is flowing into fundamentally different approaches to intelligence. Whether you’re building, buying, or just trying to keep up — this is the kind of week that defines 2026 as the year AI goes truly mainstream.

Sources

• Anthropic capturing 73% of first-time enterprise AI spend — Sherwood News
• Anthropic turns the tables on OpenAI in critical revenue category — Axios
• Anthropic nears $20B revenue run rate — Bloomberg
• Introducing GPT-5.4 — OpenAI
• OpenAI launches GPT-5.4 with Pro and Thinking versions — TechCrunch
• Introducing GPT-5.4 mini and nano — OpenAI
• Gemini 3.1 Flash-Lite: Our most cost-effective AI model yet — Google Blog
• Google releases Gemini 3.1 Flash Lite at 1/8th the cost of Pro — VentureBeat
• Yann LeCun’s AMI Labs raises $1.03B — TechCrunch
• Yann LeCun’s AMI Labs raises $1bn in Europe’s biggest seed round — Sifted
• Four MTIA Chips in Two Years — Meta AI Blog
• Meta rolls out in-house AI chips — CNBC
• Microsoft weighs suing OpenAI over Amazon cloud deal — Winbuzzer
• Microsoft considers legal action over $50bn contract — DCD