Qualcomm Accuses Apple of Stealing Trade Secrets and Giving Them To Intel (betanews.com)
Mark Wilson writes: Chip-maker Qualcomm has today accused Apple of stealing trade secrets and sharing them with Intel. The company alleges that Apple wanted Intel to be able to improve its own chips so it could move away from using Qualcomm's. Qualcomm and Apple are already engaged in a legal battle, and with its latest accusations, the chip-maker wants the court to amend its existing lawsuit against the company. Apple stands accused of engaging in a 'multi-year campaign of sloppy, inappropriate and deceitful conduct'. In the new filings, Qualcomm says that upon Apple's request it allowed the iPhone maker deep access to its software and tools, but with strict limits on how those products could be used. It said, "Indeed, it is now apparent Apple engaged in a years-long campaign of false promises, stealth and subterfuge designed to steal Qualcomm's confidential information and trade secrets for the purpose of improving the performance of lower-quality modem chipsets, with the ultimate goal of eliminating Qualcomm's Apple-based business."
That's what gives the allegation a veneer of credibility. Apple has done exactly this (albeit legally) with Samsung - going with slower Toshiba NAND and SK Hynix RAM in their devices instead of Samsung. Trying to have LG manufacture the OLED panels for their iPhones as an alternative to Samsung. Apple's MO among suppliers is like Walmart - use its market dominance to bully suppliers into accepting extremely thin margins ("Well if you won't sign this hundred million dollar contract with us at a lower unit price, then we'll just ink a deal with your competitor instead."). That's partly how they maintain such a huge profit margin (over 20%, vs about 5% for the rest of the computer industry), not just by overcharging customers. But it's a tactic which doesn't work when only one supplier dominates the market.
x86 is CISC. ARM is RISC. CISC vs RISC has played out numerous times since the 1980s. CISC has won every time. I like the idea behind RISC, but its advantages don't seem to beat out the advantages of CISC when it comes to optimized processor performance. RISC seems to have an advantage when rapidly transitioning to new processing regimes (previously high-end data processing with MIPS, currently with low-power processing with ARM). But it's like once the transition has been made or slows down, CISC allows you to optimize it further, resulting in CISC beating out RISC long-term.