Pulse 250523: Altman + Ive, Qatar's Quantum Leap, and Not Trusting Anyone Over 30
Every Friday, I share 3 takes on tech, geopolitics, infrastructure, or AI; 2 interesting reads from others; and 1 soundtrack to wind down for the weekend.
3 Quick takes from me
I. Full stack
On Wednesday, OpenAI announced it would acquire the remaining 77% of Sir Jony Ive’s hardware startup, io, for a whopping USD 6.5 billion. The 55-employee team will join OpenAI, while Ive will “assume deep creative and design responsibilities” through his independent design firm, LoveForm. OpenAI already held ~23% of io. Buying the rest at a ~$6.5B valuation implies a massive premium for a 55-person consultancy.
Needless to say, this is a promising moment for AI consumer hardware—amid longstanding doubts about GPT products’ commercial viability—and another testament to OpenAI’s incredibly fast execution on both short- and long-term strategy. An internal draft of ChatGPT’s H1 2025 strategy, surfaced via U.S. v. Google (2024) (1:20-cv-03010), helps illustrate how this acquisition underpins virtually every major long-term goal and ambition.
From bot to superassistant
OpenAI seeks to evolve GPT into a super assistant “that knows and cares about its user” and “can help with any task that a smart, trustworthy, and emotionally intelligent person with a computer could do.”
Frontier reasoning models may be mature for agentic workflows, but delivery remains a challenge. Chatbot UIs alone are too constrained to deliver on the promise and experience of agentic workflows at the core of a super assistant. As chatbot fatigue grows among users, “embodiment” has become the new frontier: hardware must balance usability and seamless integration into day-to-day life, emotional resonance, and user adoption.
Voice has more psychological resonance than visual UIs, which Altman has openly tied to OpenAI’s future, despite popular voice assistants such as Alexa and Siri generally failing to deliver optimal user experiences.
Only the best
“We need the best free model, best UI, and strongest brand.” What they’re buying is Ive’s vision and myth behind the iPhone, iPod, and iMac, products whose visual appeal, seamless integration of hardware with software, and emotional resonance turned them into objects of desire. OpenAI doesn’t want to build the next iPhone, but rather a “category-defining product” to outshine competitors who already leverage vertical integration in their hardware products as an embedded distribution channel for AI.
New moats
Growth and revenue won’t line up forever.” DeepSeek-R1’s release showed that open weights and lower compute requirements reduce the barrier to entry and allow anyone to build on open-source foundational models—and capture market share in more defensible layers of the value chain, like agent architectures, orchestration, applications, and services. Defensible revenue moats are shrinking. OpenAI is trying to diversify across other layers of the value chain to preserve its competitive advantage.
DAUs and demand
The H1 strategy notes the need to build demand for monetizable use cases (e.g., search, planning, scheduling, transactions). A differentiated device would increase daily touchpoints (e.g., through voice), enable higher-margin, premium GPT experience, and potentially integrate first-party monetization models (vs. App Store developer tolls, which Apple is currently being litigated against). Building and integrating hardware is about locking users into ecosystem economics.
But all this doesn’t come without being conscious of risk vs. burn rate. Since 2019, OpenAI has secured USD 13bn from Microsoft and is currently negotiating additional investment to fuel a future IPO. With AGI still years away, OpenAI needs near-term revenue, and hardware is famously low-margin (unless it becomes a recurring-revenue driver as mentioned above):
Supply chains, returns, margins, and retail are all high CAPEX/OPEX activities and introduce significant overhead, which OpenAI has avoided so far due to its lean supply chain infrastructure, and in part by its lean headcount (2000+ employees as of writing).
A standalone device might justify a luxury price point at risk of being too expensive or niche, as seen with VR/AR devices like Apple Vision Pro (USD 3,499).
A companion device (i.e., requiring a paired smartphone, like an Apple Watch) might reduce cost, expand the serviceable addressable market, and deliver deep integration (but without owning the OS layer).
The first product is expected to debut in 2026.
II. [Update] Commerce Doubles Down
As I suggested last week, the U.S. Commerce Department just placed subsidiaries of Chinese firms such as CXMT and SMIC on the Bureau of Industry and Security’s (BIS) Entity List. Should these go into effect, Commerce would be putting China and the global semiconductor stack in a chokehold.
For instance, Brite, a fabless subsidiary, is tightly coupled to SMIC and relies on IP from the U.S. and allied nations (e.g., Germany, Israel, Singapore). Meanwhile, key players in those countries are now further exposed to sanctions should they unknowingly inherit hardware, software, or technology with controlled components from Chinese suppliers.
BIS is escalating to target upstream points (fabs, foundries) to maximize downstream impacts throughout the global AI supply chain. U.S. firms must now have to justify and audit every downstream node of their chain and product usages. This escalation is unfolding amid a broader tariff war détente, the timing of which is consistent with the White House’s often inconsistent strategies when it comes to China (despite an overall hawkish cabinet).
Historically, Chinese firms responsible for the country’s military modernization have persisted in sourcing U.S.-origin technology and equipment even if their parent company was blacklisted. This announcement, if finalized and enforced, may do more than merely isolate China: an active attempt at rewiring the global tech stack via export controls, sanctions, and other non-tariff barriers that may compel allies to align with the U.S. However, it assumes compliance can scale and be enforced faster than China’s ability to circumvent controls (yet the DeepSeek R1 release showed controls hardly keep up with diffusion velocity).
III. Quantinuum goes Qatari
Quantinuum, a leading U.S. quantum computing company backed by Honeywell, recently secured a substantial USD 1bn from Al Rabban Capital. This 10-year partnership aims to accelerate quantum technology adoption in the Middle East, notable applications of which span finance, medicine, and genomics. While this deal underscores the global appetite for quantum advancements, Quantinuum finds itself entangled in highly intricate geopolitical tensions in the region.
Quantinuum will be able to diversify its client base beyond traditional Western defense and enterprise clients by tapping into regional markets. However, Qatar’s strategic relationships with both Western nations and China, Russia, and regional, U.S.-sanctioned actors (Syria, Iran, Hamas) may prompt scrutiny from the U.S. (and potentially the EU), notably because their products have dual-use capabilities.
For instance, its flagship product, a quantum-enhanced cryptographic key generation platform for cybersecurity, is prone to dual-use regulation for its use of randomness techniques and complex protocols and potential for integration in cyber intrusion tools and other military systems. Such techniques and protocols have historically fallen under Category 5, Part 2 (Information Security) under the U.S. Export Administration Regulations (EAR) and the EU Dual-Use Regulation—among the most stringent categories of either list.
As mentioned above—and despite the significant infrastructure deals struck during Donald Trump’s Middle Eastern tour last week—a caveat remains: Qatar, Saudi Arabia, and the UAE continue to position themselves as non-aligned actors in the global AI competition between the U.S. and China. One should not assume these countries won’t also court China on quantum, particularly as Qatar seeks to present itself as an attractive destination for deep tech investments.
2 Compelling reads from others
I. Unity & Humility
Pope Leo XIV ascended to the papacy last Sunday, calling for peace, unity, and humility, declaring he “would not rule as an autocrat, but to serve with humility and compassion.”
II. Can ads scale with agents?
Another forward-looking piece from Stratechery arguing that while a necessary economic model for the “human web,” advertising is fundamentally incompatible with an “agentic web” where AI, not people, consume content. I highly recommend subscribing (free or paid) if you like systems thinking and sharp takes on tech operations.
1 Musical Note
Claudya’s Com Mais de 30 is a striking response to the countercultural mantra: “Don’t trust anyone over 30.” Released in 1971, the song presents turning 30 not as a loss of relevance or authenticity, but as a moment of clarity, complexity, and fresh perspectives.
Thanks for reading SecondOrder! Subscribe for free to receive new analyses. If you know someone who might enjoy this content, please forward this email.
Have thoughts on my takes? Comment on Substack or send me a message!