Apple (AAPL) on Solana

U.S. lawmakers are pressing the Commerce Department to bar Apple from purchasing memory chips from Chinese supplier CXMT, which carries a "Chinese Military Company" designation. The central argument is that every chip Apple buys from CXMT effectively subsidizes the People's Liberation Army, raising national security concerns that extend beyond Apple itself to the long-term dependence on Chinese memory in AI systems and critical infrastructure.

Memory chips are a core input for iPhones, Macs, and Apple's AI-focused hardware lineup, so a successful sourcing ban would force the company toward alternative suppliers — Samsung, SK Hynix, and Micron — potentially at higher cost and with capacity constraints. The push reflects the broader US-China tech decoupling trend, where policymakers are increasingly targeting component-level supply chains rather than finished goods, putting pressure on major platform companies to audit and restructure deep-tier supplier relationships.

China's Cyberspace Administration has formally approved Apple Intelligence for deployment in the country, with Apple integrating Alibaba's Qwen AI model to power generative features — including text and image generation — across iOS, iPadOS, macOS, and visionOS for Chinese users. The clearance resolves months of regulatory uncertainty around how Apple would bring its flagship AI software to its largest overseas market, where domestic rivals like Huawei have been aggressively marketing their own AI-driven devices.

Apple shares jumped roughly 4.2% to a record $328.53 on the news. To make the experience viable on-device, Apple is in talks with PrismML, a Silicon Valley startup that compressed Alibaba's 27-billion-parameter Qwen model from approximately 54 gigabytes to under 4 gigabytes — small enough to run natively on an iPhone 15 without relying on cloud processing. Evercore ISI reiterated an Outperform rating and $365 price target following the approval.

Apple is in early-stage talks with PrismML, a Khosla Ventures-backed startup spun out of Caltech, about technology that drastically shrinks the size of large AI models so they can run directly on iPhones. PrismML's approach reduces the numerical precision of a model's internal values from 16 bits down to just one or three, allowing a 27-billion-parameter model that normally occupies 54 gigabytes to be compressed to under 4 gigabytes — well within the 8GB of memory in the iPhone 15 Pro and Pro Max. The company recently published compressed versions of Alibaba's open-source Qwen model to demonstrate the technique.

PrismML CEO Babak Hassibi confirmed to CNBC that Apple and other companies have been testing the startup's models for speed, energy efficiency, and on-device performance, characterizing the discussions as very early but noting that "things are progressing nicely." For Apple, the technology could allow significantly more capable AI to run locally rather than in the cloud, supporting the company's emphasis on privacy and potentially enabling a more powerful Siri without relying on server-side inference.

Apple's lawsuit against OpenAI, filed July 10, drew immediate public commentary from Elon Musk on X, where he called Sam Altman "Scam Altman" and accused him of graduating from "stealing an open source AI charity" to stealing Apple's phone technology. Altman fired back, mocking Musk's orbital data center pitch and suggesting the attention itself confirmed OpenAI's leading position in AI. OpenAI responded formally that it has "no interest in other companies' trade secrets" and is focused on building its own technology.

Beyond the social media exchange, the lawsuit surfaces new detail about the alleged misappropriation. Apple names not only Tang Tan — OpenAI's chief hardware officer and former Apple VP of product design who allegedly emailed himself supplier information and encouraged Apple employees to bring confidential hardware to job interviews — but also Chang Liu, a technical staff member who allegedly downloaded confidential Apple files and coached a colleague to copy additional materials. The breadth of the complaint points to Apple treating its AI hardware supply chain and chip design roadmap as core trade secrets, and signals the company intends to assert legal boundaries aggressively as its former executives migrate to competing AI ventures.

Apple filed a federal lawsuit against OpenAI in Northern California on July 10, 2026, alleging a coordinated scheme to steal Apple's AI hardware trade secrets that reached "at every level" of the company — from technical staff to OpenAI's Chief Hardware Officer. The complaint centers on Tang Tan, a former Apple VP who became OpenAI's hardware chief, and alleges he directed Apple employees interviewing at OpenAI to bring "actual parts" from Apple for "show and tell" sessions. Apple also claims OpenAI coached departing employees on how to evade Apple's security processes and that one former employee, Chang Liu, stole an Apple laptop on the way out. A manufacturing partner was allegedly manipulated into revealing a proprietary metal-finishing technique under the false impression Apple had authorized the disclosure.

The lawsuit marks a dramatic rupture in what had been a high-profile partnership: Apple integrated ChatGPT into iOS beginning in 2024. For Apple, the litigation signals how seriously the company treats the secrecy of its custom silicon and hardware manufacturing processes — areas where its supply-chain and fabrication innovations are central to its product differentiation strategy. A successful claim could result in injunctive relief limiting OpenAI's use of any derived knowledge, as well as damages. OpenAI denied the allegations, stating it has "no interest in other companies' trade secrets."

Apple has announced a more than $30 billion commitment to Broadcom to produce advanced chips on U.S. soil, marking the largest single component of Apple's previously announced $600 billion, four-year American investment plan. The deal will result in more than 15 billion U.S.-made chips, with Broadcom expanding its Fort Collins, Colorado facility by $1.5 billion to manufacture advanced radio frequency components — including FBAR filters — and wireless connectivity technologies used in Apple devices.

The agreement, which extends Apple's partnership with Broadcom through 2031, is the largest commitment under Apple's American Manufacturing Program (AMP), an initiative aimed at expanding domestic supply chain production and reducing dependence on overseas semiconductor sources. The announcement comes as Apple faces continued pressure from the Trump administration to bring manufacturing activity back to the United States.

Between 2013 and 2026, Apple directed approximately $853 billion into share repurchases — a sum large enough to have acquired any of 487 companies currently in the S&P 500. Rather than pursuing a transformative acquisition, CEO Tim Cook and Apple's board consistently chose to buy back their own stock, retiring more than 44% of outstanding shares over that period. Annual repurchases scaled from $22.95 billion in 2013 to a peak of nearly $95 billion in 2024, with the 2022–2024 stretch seeing buybacks consistently above $77 billion per year.

The strategy reflects a deliberate capital-return philosophy built on three pillars: buybacks mechanically lift earnings per share by shrinking the share count, the 2017 Tax Cuts and Jobs Act's reduction of the corporate tax rate from 35% to 21% freed up more retained earnings for repurchases, and steady buyback programs tend to attract long-term shareholders while dampening volatility. Critics note the approach prioritizes financial engineering over growth through acquisition, but Apple's consistent execution has made its buyback program one of the largest in corporate history.

Apple is targeting production of roughly 10 million units of its first foldable iPhone, with the device expected to ship between H2 2026 and H1 2027 as part of a lineup that now includes at least five new iPhone models. The company is also introducing an iPhone Ultra positioned at the top of the product range, widening the portfolio beyond the existing Pro Max tier.

Alongside the hardware expansion, Apple is reportedly diversifying its memory chip supply chain toward Chinese suppliers that are on the Pentagon's blacklist, accepting political and reputational risk in exchange for supply security. The foldable launch volume and Ultra pricing strategy will be closely watched by investors given that AAPL shares are currently trading above internal fair-value estimates, according to analyst commentary cited in the report.

Apple is lobbying the US government for permission to purchase memory chips from ChangXin Memory Technologies (CXMT), a Chinese manufacturer currently on the Pentagon's blacklist, according to supply chain analyst Ming-Chi Kuo and Financial Times reporting. The move follows Apple's recent round of price hikes on MacBooks, iPads, HomePods, and Apple TV devices, which the company attributed to surging memory and storage costs driven by AI data center demand.

Kuo cautioned that even if the lobbying effort succeeds, relief may be limited: CXMT's production capacity is heavily absorbed by domestic Chinese demand, meaning it would not substantially close the supply gap or reduce costs for Apple. He identified a deeper structural issue — memory capacity is being diverted toward AI infrastructure, tightening supplies for consumer electronics broadly and putting iPhone production levels at potential risk. AAPL rose 0.4% in overnight trading ahead of Monday's open after falling 6.1% the prior week when the price increases were announced.

Apple is overhauling its Mac chip roadmap by canceling the M6 Pro and M6 Max variants and moving directly to an M7 generation purpose-built for AI workloads, according to Bloomberg's Mark Gurman. The company will still release a base M6 chip for entry-level Macs as early as late 2026, but professional-tier buyers will wait for the M7 line—a first in Apple Silicon history where a chip generation ships without a Pro or Max variant.

The M7 is designed with expanded on-device AI capabilities and a significantly faster GPU, with memory bandwidth reportedly rising to 240 GB/s from the M6's 200 GB/s. Apple plans to begin rolling out M7 chips in the first half of 2027, followed by M7 Pro and M7 Max models by year-end and an M7 Ultra in 2028, accelerating the availability of its most capable silicon for users running AI-intensive software.