The price action on Sivers Photonics over the past fortnight whispers a story the order books don't yet confirm. The stock surged 40% on rumors alone—a classic retail chase into a binary event. But I've seen this pattern before. In 2021, during the DeFi summer, the same herd mentality drove liquidity pools to unsustainable APYs. Today, it's compound semiconductor laser stocks. The catalyst? AMD's upcoming 'Advancing AI' event on July 22-23, where the company is expected to unveil its MI500 GPU with co-packaged optics (CPO). The market is pricing in a revolution. But revolutions leave many ghost towns.
Here's the context: AMD, the perennial challenger to NVIDIA in the AI GPU race, is betting its next-generation MI500 on a radical interconnect architecture. Traditional electrical traces in the rack can no longer handle the bandwidth demands of GPT-scale training clusters. CPO—embedding optical engines directly next to compute dies—promises to bypass that bottleneck. AMD is reportedly collaborating with GlobalFoundries (GF) and Ayar Labs to make this happen. The key upstream component? Lasers. And that's where Sivers Photonics enters the narrative, as a supplier of indium phosphide (InP) laser diodes, indirectly feeding into GF's silicon photonics platform.
Core Insight: The Option Pricing of a Supply Chain Role Over the past six months, I've audited over a dozen semiconductor supply chain models for mid-cap firms. What I see with Sivers is a textbook case of 'option pricing' in the equity market. Investors are buying a call option on AMD's CPO roadmap—an option that expires on July 22. If AMD explicitly mentions Sivers as a reference design partner, the stock could double overnight. But if the event is vague, or if AMD chooses Lumentum or Coherent as the volume supplier, the premium evaporates. The market is ignoring the 40-50% probability that Sivers remains a validation partner, not a production one. My analysis of GF's SCALE platform shows that while Sivers' laser design may be in the reference library, commercial customers demand proven reliability at scale. InP lasers are notoriously difficult to manufacture with high yield—a challenge that favors incumbents with decades of experience.
Contrarian Angle: The Hype Ignores the Real Risk — and the Real Opportunity The consensus narrative is binary: AMD success → Sivers success. But that's a dangerous oversimplification. Let me offer three contrarian observations based on my work with institutional asset managers entering crypto hardware plays. First, AMD's route to volume is through GF's fab in New York, not through a single laser vendor. GF may dual-source to mitigate risk, completely diluting Sivers' share. Second, the CPO itself won't hit mass production until 2026—meaning the revenue timeline is two years out, yet the stock already trades at 12x forward sales. Third, and most importantly, the real value creation in CPO may not be in the laser itself but in the integrated photonic engine. Companies like Ayar Labs (private) or Broadcom hold the system-level patents. Sivers in this context is a component supplier with weak bargaining power. The market is pricing it as a monopoly, but the reality is more fragile.
I've been through this before—in 2022, when I retreated to the Mekong Delta after the bear market wiped 40% of my portfolio. I spent three months building a Python simulator for ZKP trading strategies. That taught me to value second-order effects over first-order narratives. The first-order narrative is Sivers wins. The second-order is that CPO adoption will force a radical restructuring of the optical supply chain, but the winners will be firms with diversified revenue, not single-customer bets.
Takeaway: Watch the signals, not the noise Over the next 72 hours, focus on three things: Whether AMD's PR mentions 'reference design' vs. 'production partner'; whether Jefferies or Morgan Stanley publishes a note citing Sivers; and the CEO's body language on X. The ledger remembers what the market forgets—that hype is a prelude to repositioning, not a conclusion. If Sivers gets the nod, ride the momentum but set a hard stop at 20% from the peak. If not, the ghost trade will be painful.
Between the block and the breath, truth resides. This time, the block is silicon photonics; the breath is the laser's coherence. Trade accordingly.
