Peak Memory Fears Put Focus on Samsung, SK Hynix Cash Returns
THE SO WHAT
When memory vendors start talking more about cash returns than capex, the market is bracing for a plateau in AI-driven demand. If you’re building long-term around cheap, ever-expanding HBM and DRAM, start modeling scenarios where pricing power returns and supply growth slows.
READ THE SOURCE
MORE FROM THE WIRE
Tech & InnovationChina Earnings to Test Rotation From AI to Internet Stocks
The fact that investors are even debating a rotation from AI infra into China’s big internet names tells you AI is starting to be treated as a cyclical trade, not a one-way bet. If you’re selling into either side of that barbell—chips or consumer internet—expect capital availability and valuation multiples to swing on quarterly prints, not long-term narratives.
Tech & InnovationCapturing the 'primal energy of Wolverine' in a new look, Sony is going all-in on Marvel’s Wolverine with a new PS5 collection that includes covers, a DualSense, and a limited-edition console
Sony keeps proving that IP-driven hardware skins are a durable monetization layer on a flat console base. If you run a platform with strong franchises, the playbook is clear—treat hardware cosmetics as recurring revenue, not marketing swag.
Tech & InnovationSource: Kalshi's annualized revenue topped $4B in July, up from $2B+ two months earlier, and its operating expenses totaled $300M in June
A venue doubling to a $4B+ annualized run rate on World Cup flow while running $300M in monthly opex shows how fast real-money prediction markets can scale once they hit cultural events. If you’re in consumer finance or trading, assume these markets will become a mainstream acquisition and engagement surface, not a niche side bet.
Tech & InnovationRedpoint’s Brescia on Nvidia’s $500B Funding, AI Capex
When VCs are talking about $500B+ AI capex as a baseline, the message is that infra spend is behaving like a new utility build-out, not a cyclical IT upgrade. If you sell into this stack, orient roadmaps and contracts around multi-year capacity planning, not quarter-to-quarter GPU scarcity.