Lumentum reports surging AI component demand with orders booked through 2028

Lumentum Holdings Inc., backed by Nvidia, reported surging demand from U.S. hyperscalers for its optical components, with AI-related orders expected to fill its books through 2028. According to the company, this acceleration underscores the unrelenting appetite for infrastructure to power artificial intelligence applications.^[3]

The announcement comes amid a broader frenzy in AI hardware and cloud computing, where major tech firms are pouring billions into capacity to meet explosive demand. Lumentum's outlook highlights how even niche suppliers in the optical transceiver space—critical for high-speed data transfer in data centers—are benefiting from the AI boom. Hyperscalers like Meta, Amazon, and others are racing to scale up, driving orders that extend years into the future and signaling no near-term slowdown despite geopolitical tensions.^[3]

CoreWeave Inc., a key AI cloud provider, exemplified this trend by securing a massive $21 billion commitment from Meta Platforms Inc. for dedicated AI computing power, bringing their total partnership to about $35 billion through 2032. The deal includes early access to Nvidia's next-generation Vera Rubin platform across multiple sites, prompting CoreWeave to tap the junk debt market for funding after previous borrowing sprees.^[1][4] Bloomberg Technology noted that this underscores Meta's aggressive push into AI infrastructure.

Taiwan Semiconductor Manufacturing Co. (TSMC), the world's leading chip foundry, further validated the sector's resilience with a 35% quarterly revenue jump that exceeded estimates. As reported by Bloomberg Markets, global AI chip demand held firm even in the initial weeks of conflict in the Middle East, showing how AI priorities are overriding external shocks.^[2]

Competition is intensifying as cloud giants develop their own chips to challenge Nvidia's dominance. Amazon CEO Andy Jassy revealed in his annual shareholder letter that the company's custom silicon business—including Graviton processors, Trainium AI chips, and Nitro—now runs at over $20 billion in annualized revenue, growing at triple-digit rates. Jassy hinted at selling Trainium chips directly to third parties, potentially valuing the unit at $50 billion if opened to the broader market, a move that could erode Nvidia's lead.^[5] Additional reporting from TechCrunch and The Information confirms Amazon's Trainium chips are attracting major AI players like OpenAI and Anthropic, with commitments for gigawatts of capacity and performance claims up to 50% cheaper than Nvidia GPUs for certain tasks.^[1][5]

Oracle, meanwhile, appointed Hilary Maxson as its new CFO to oversee a $50 billion capital expenditure surge for AI data centers. The former Schneider Electric executive steps in at a pivotal moment, reporting to CEO Clay Magouyrk as Oracle ramps up to support its AI ambitions.

These developments matter for investors, tech workers, and the global economy, as AI infrastructure spending—projected in the hundreds of billions—fuels job creation in semiconductors and data centers while raising questions about supply chain bottlenecks and energy demands. What happens next includes potential external chip sales from Amazon, CoreWeave's debt-fueled expansion, and ongoing hyperscaler deals that could lock in capacity through the decade, per Lumentum's visibility.^[3][5] Stakeholders from chipmakers to cloud providers remain poised for sustained growth, even as rivals like Google and startups vie for share in this high-stakes race.