Astraas·Author

Raunak A.

Writes about technology for Astraas — hardware, software, AI and the money moving through all three.

• 23 Aug GrapheneOS is escaping Pixel exclusivity. The price is the catch GrapheneOS will support Motorola flagship phones from 2027, priced above the Pixel lineup. Cheaper devices will take…
• 23 Aug Walmart's decade of refusing Apple Pay ends Monday, in Arkansas Walmart starts accepting Apple Pay, Google Pay and contactless cards on 24 August in Arkansas, expanding state by sta…
• 23 Aug The GTA 6 leaks have a name now. Take-Two wants the address Take-Two has petitioned a New York court to subpoena Microsoft and Discord for the identities behind the GTA 6 leaks:…
• 22 Aug The 200 jobs Apple cut are next year's product roadmap Apple is shedding some 200 roles from its headset and assistant organisations. The Vision Pro gaming team is being sh…
• 22 Aug TikTok's $400 million settlement has a $100 million asterisk The DOJ's 2024 COPPA suit against TikTok ends with a $400 million payment, a quarter of it conditional on unwinding a…
• 22 Aug A million people have reported AI-sounding posts on LinkedIn. Its writing tools helped write them LinkedIn says a million people have clicked its AI slop button and views on classified slop are down 40 percent. Pang…
• 7 Aug The GTA 6 deal makes more sense once you see which Netflix desk it came from Netflix gets six hours with the next Grand Theft Auto 6 footage before it reaches YouTube on 27 August. The Netflix e…
• 7 Aug Google built an agent that does everything except pay Ask Maps can now order food, book hotels and plan days across 150 more countries. Google says it completes every step…
• 7 Aug The Private Relay leak matters because no iOS browser can route around it Tommy Mysk and Talal Haj Bakry found that WebKit's passkey handling bypasses iCloud Private Relay, exposing a real IP…
• 4 Aug The subscription breakeven is 154 million tokens a month At published API rates, $100 buys roughly 154 million tokens on a cache-heavy agentic workload. That is the point whe…
• 4 Aug A cache entry pays for itself on the second read Cache writes cost 1.25x and cache reads cost 0.1x, which puts the break-even at 1.39 reads. Measured across 9.75 bill…
• 4 Aug Routing 7% of calls to cheap models saved 0.89% Routing 6.9% of calls to Haiku and Sonnet cut 0.89% of spend. Measured across 23,157 calls, cheap-model routing is a…
• 4 Aug A backtest said Sharpe 2.74. Real data said 0.35 Nineteen backtested strategies, and the headline finding is not the fourteen failures. It is that the best proxy resu…
• 4 Aug Inference got 95% cheaper. Your bill went up anyway Per-token inference prices have collapsed — 95% in two years by some measures — while enterprise LLM spend has double…
• 4 Aug 95% of enterprise AI pilots return nothing. The failure isn't the model MIT found 95% of enterprise GenAI pilots produce no measurable P&L impact against $2.52 trillion of AI spending. The…
• 3 Aug 9.96 billion tokens in 34 days. Output was 8% of the bill Thirty-four days of real Claude Code usage, costed line by line at published API rates. Cache reads were 98% of token…