Nvidia $NVDA is now worth more than India 🇮🇳
Source: Bloomberg, Barchart
One of the biggest fears on Wall Street right now? That the surge in IPOs and equity issuance could trigger a market sell-off.
But history says the opposite. According to Deutsche Bank’s Jim Reid and strategists Binky Chadha & Parag Thatte, issuance waves usually happen because markets are strong — not because they’re about to crash. Companies raise capital when: • Investor demand is high • Earnings momentum is strong • Risk appetite is elevated Since early 2023, US equity issuance has jumped from ~$30bn to ~$120bn per quarter. Mega-IPOs are coming. Yet even the largest deals are only ~0.1% of the S&P 500 market cap. Past issuance cycles? Median returns were: • +8% over 3 months • +20% over 12 months The only major exception: 2008. Bottom line: strong demand is absorbing new supply. This market still feels a lot more like 1999 than 2008. Source: Zerohedge, DB
A positive contrarian note on SpaceX and Anthropic Saying they have overvalued seems to be the consensus
A positive contrarian note on SpaceX and Anthropic Saying they have overvalued seems to be the consensus
S&P 500 ex AI vs. S&P 500 5-day change
That is the biggest spread since AI was birthed... Source: zerohedge, Bloomberg
Oil falls. Yields rise. Something has shifted
Oil down 15% in three weeks. The two-year yield up 15bps last week to 4.15% - its highest since early 2025. For the first time in months, US macro is back in the driving seat. Source: Jonny Matthews | SuperMacro
Rotational market lives on. We note that 368 stocks were higher in the S&P 500 yesterday. The most since late April.
Rare to see so many stocks going up with the S&P 500 retreating. Source: Ryan Detrick
From Mag7 to MANGOS
Source: Evan
This chart is comparing how different AI models perform on a set of benchmark tests — mainly around coding, reasoning, and knowledge tasks.
The key takeaway: Claude Mythos 5 / Fable 5 is claiming better performance than GPT-5.5 and Gemini 3.1 Pro on these specific benchmarks. Example: SWE-Bench Pro (agentic coding): Claude: 80.3% GPT-5.5: 58.6% Gemini 3.1 Pro: 54.2% What does that actually mean? “Agentic coding” This measures how well an AI can: understand a software engineering task, navigate codebases, edit files, debug issues, and complete coding workflows autonomously. So higher % = the model solved more real-world coding tasks correctly.
