Korean AI startup Upstage says it is in talks to acquire 10K of AMD’s MI355 chips, in a bid to “diversify to other chips” as “we have a lot of Nvidia chips”
https://www.bloomberg.com/news/articles/2026-03-23/ai-startup-upstage-looking-at-buying-10-000-amd-chips-in-korea
Upstage’s push to diversify beyond Nvidia is less a strategic masterstroke and more a tacit admission of the risks inherent in over-reliance on a single dominant supplier. AMD’s MI355 chips, while promising, lag behind Nvidia’s entrenched ecosystem and software maturity. This move highlights the underlying fragility of AI hardware supply chains where startups scramble for alternatives—not out of preference but necessity amid geopolitical tensions and supply constraints. Diversification isn’t innovation; it’s a hedge against a monopolistic chokehold.
A look at Huawei-backed Yuanjie, a maker of photonic chips used in AI data center optical interconnects, whose stock has surged 780% over the past year
https://www.forbes.com/sites/ywang/2026/03/20/ai-optics-boom-propels-founder-of-photonic-chip-maker-into-the-billionaire-ranks/
Yuanjie’s staggering stock surge is a classic example of how geopolitical alignment inflates valuations detached from global market realities. Huawei’s deep involvement signals state-driven strategic positioning, not purely market-driven innovation. Photonic chips will be critical for data centers, but Yuanjie’s meteoric rise masks the risk of decoupling from Western technology ecosystems. Investors chasing the optics boom risk being trapped in a China-centric supply chain increasingly isolated by export controls and trust deficits.
OpenAI hires Dave Dugan, a former top ad executive at Meta, to lead ad sales, reporting to COO Brad Lightcap; Dugan stepped down as Meta VP earlier this month
https://www.wsj.com/tech/ai/openai-taps-former-meta-executive-to-lead-ad-push-60d39af2?st=Agy4yE&reflink=desktopwebshare_permalink
OpenAI’s pivot to ad sales under a former Meta ad chief signals retreat from its pure AI research roots toward replicating the very data-exploitative ad model it once seemed poised to disrupt. This move risks eroding user trust and undercuts claims of ethical AI leadership. The real story is OpenAI’s acceptance that sustainable revenue emerges from the entrenched, invasive surveillance economy—not breakthrough AI products or enterprise solutions.
Sources: OpenAI is offering PE firms a guaranteed minimum return of 17.5% and early access to new models to secure JVs, an improvement on Anthropic’s terms
https://www.reuters.com/business/openai-sweetens-private-equity-pitch-amid-enterprise-turf-war-with-anthropic-2026-03-23/
OpenAI’s sweetened private equity offerings reveal cracks in its business model and growing dependency on financial engineering to sustain growth. Guaranteeing hefty returns and early model access is a red flag: it signals desperation to lock in funding amid intensifying competition with Anthropic and others. This underscores the precariousness of the AI gold rush, where the race for capital might overshadow product viability or ethical considerations. Guaranteed returns hardly align with genuine innovation risk-taking.
3 men are charged with conspiring to smuggle US artificial intelligence to China
https://apnews.com/article/artificial-intelligence-china-charges-e8f5135a71b8863c66b9c73d04cb0eb2
The indictment of individuals smuggling US AI to China is a microcosm of a far broader, systemic leakage that no export control regime can fully contain. Behind these arrests lies a sprawling, covert network fueled by demand for cutting-edge tech and the incentives to circumvent restrictions. This case exposes the futility of current US policy frameworks that focus on symptomatic arrests without addressing root causes like globalized supply chains, talent flows, and the inherent difficulty of policing intangible assets in AI.
Nvidia CEO heralds ‘inference inflection’ as next phase of AI boom, backed by $1 trillion in orders
https://apnews.com/article/nvidia-ceo-jensen-huang-artificial-intelligence-conference-846f7d4aada068e92516665c6993ea29
Nvidia’s trillion-dollar order backlog and talk of an “inference inflection” mask the unsustainable hype cycle driving AI hardware demand. The staggering figures reflect speculative spending more than proven, profitable use cases at scale. Nvidia’s dominance risks creating a monoculture vulnerable to supply shocks, regulatory pushback, or technological disruption. This inflection point is less a sign of maturity and more a warning that AI’s hardware bottleneck is consolidating power dangerously in one player—inviting geopolitical and market turbulence ahead.
Sources: Hacker News, Techmeme, AP News, Ars Technica | Compiled 2026-03-23