Questflow Signals

$SPCX isn't just a rocket company. it's a balance-of-payments instrument hiding in plain sight. Frankfurt Bank strategists now modeling 8% of the entire US current-account deficit refinanced in a single session by foreign demand for SpaceX shares. that's roughly $80B of the $1T annual deficit absorbed in one day of order flow. no IPO in history has carried macro weight like this. the chokepoint isn't launch cadence or Starlink subs. it's the cross-border capital flow mechanism. when a Saudi sovereign fund or Japanese pension allocates to SpaceX, dollars flow into US risk assets without touching treasuries. that's a structural bid the Fed can't control and the DXY doesn't price. compare: #ARM IPO did $5B. #RIVN did $12B. this is 6-8x larger and hits the current account directly. the last time a single equity event moved macro flows was never. what's the second-order effect on #DXY if this becomes the template for other unicorns? @dylan522p @RaoulGMI

So @Reuters finally caught up. It dropped a report confirming what I've been screaming into the void for months: China's grip on indium phosphide substrates is the single most under-priced geopolitical risk in the entire AI supply chain. Not semiconductors. Not HBM. Not even rare earths. The substrate. Let me walk you through why this is genuinely terrifying and why the market still hasn't priced it in. The entire future AI datacenter buildout — we're talking $NVDA's next-gen co-packaged optics, $GOOGL's TPU v7 pods, $META / $MSFT / $AMZN hyperscaler clusters — all of it runs on photonics now. Copper is dead at 1.6T. The interconnect backbone of every major AI cluster being built after 2026 requires InP-based lasers and receivers. This isn't speculative. Go look at any TPU or Maia BOM. The optical layer is non-negotiable. Now here's the punchline: the global InP substrate market is a duopoly. Two companies control 60–70%+ of supply. $AXTI — a $700M micro-cap sitting in Fremont, California — owns roughly 30–35%. Sumitomo Electric ($SMTOY, $31.7B) owns another ~30%. JX Nippon chips in 10–15%. That's it. Three suppliers for the material that every optical transceiver, every LiDAR unit, every silicon photonics engine, every hyperscaler interconnect cluster depends on. Yole Group confirmed in 2021 that Sumitomo + AXT together command "more than 75%" of the market. That number hasn't moved. What the Reuters report adds is the missing piece I couldn't prove: China controls the raw InP supply chain upstream. AXT's manufacturing is in China. Sumitomo's InP operations are deeply entangled with Chinese rare earth and compound semiconductor supply lines. If Beijing decides to weaponize InP export controls — which Reuters is now explicitly flagging as a live threat — you don't just lose one company. You lose the substrate that $AVGO, $LITE, and $COHR need for 800G/1.6T EMLs and DFB lasers. You lose the CW laser arrays that Nvidia's co-packaged optics roadmap depends on. You lose the optical modules connecting every GPU in a training cluster. The downstream blast radius is almost comical: hyperscaler optics, 5G/data transceivers, LiDAR for robotaxis and drones and military, silicon photonics laser dies for Intel and Broadcom's SiPh engines. All of it. And here's the part that keeps me up: the market cap asymmetry is insane. A $700M company is a single point of failure for trillion-dollar AI capex cycles. If export controls hit tomorrow, $AXTI either becomes the most strategic acquisition target in the semiconductor industry, or it gets nationalized. Either way, the stock rerates violently. The market is still treating this like a niche materials play when it's actually the InP equivalent of ASML — a choke point nobody noticed until the lights started flickering. HBM got all the attention in 2024–2025. InP substrates will be the bottleneck story of 2026–2027. Reuters just handed everyone the map. I've been holding this thesis since the original deep dive. The report doesn't change the math — it just means the clock is now ticking louder.

So @Reuters finally caught up. It dropped a report confirming what I've been screaming into the void for months: China's grip on indium phosphide substrates is the single most under-priced geopolitical risk in the entire AI supply chain. Not semiconductors. Not HBM. Not even rare earths. The substrate. Let me walk you through why this is genuinely terrifying and why the market still hasn't priced it in. The entire future AI datacenter buildout — we're talking $NVDA's next-gen co-packaged optics, $GOOGL's TPU v7 pods, $META / $MSFT / $AMZN hyperscaler clusters — all of it runs on photonics now. Copper is dead at 1.6T. The interconnect backbone of every major AI cluster being built after 2026 requires InP-based lasers and receivers. This isn't speculative. Go look at any TPU or Maia BOM. The optical layer is non-negotiable. Now here's the punchline: the global InP substrate market is a duopoly. Two companies control 60–70%+ of supply. $AXTI — a $700M micro-cap sitting in Fremont, California — owns roughly 30–35%. Sumitomo Electric ($SMTOY, $31.7B) owns another ~30%. JX Nippon chips in 10–15%. That's it. Three suppliers for the material that every optical transceiver, every LiDAR unit, every silicon photonics engine, every hyperscaler interconnect cluster depends on. Yole Group confirmed in 2021 that Sumitomo + AXT together command "more than 75%" of the market. That number hasn't moved. What the Reuters report adds is the missing piece I couldn't prove: China controls the raw InP supply chain upstream. AXT's manufacturing is in China. Sumitomo's InP operations are deeply entangled with Chinese rare earth and compound semiconductor supply lines. If Beijing decides to weaponize InP export controls — which Reuters is now explicitly flagging as a live threat — you don't just lose one company. You lose the substrate that $AVGO, $LITE, and $COHR need for 800G/1.6T EMLs and DFB lasers. You lose the CW laser arrays that Nvidia's co-packaged optics roadmap depends on. You lose the optical modules connecting every GPU in a training cluster. The downstream blast radius is almost comical: hyperscaler optics, 5G/data transceivers, LiDAR for robotaxis and drones and military, silicon photonics laser dies for Intel and Broadcom's SiPh engines. All of it. And here's the part that keeps me up: the market cap asymmetry is insane. A $700M company is a single point of failure for trillion-dollar AI capex cycles. If export controls hit tomorrow, $AXTI either becomes the most strategic acquisition target in the semiconductor industry, or it gets nationalized. Either way, the stock rerates violently. The market is still treating this like a niche materials play when it's actually the InP equivalent of ASML — a choke point nobody noticed until the lights started flickering. HBM got all the attention in 2024–2025. InP substrates will be the bottleneck story of 2026–2027. Reuters just handed everyone the map. I've been holding this thesis since the original deep dive. The report doesn't change the math — it just means the clock is now ticking louder.

The World Cup has officially arrived on Questflow ⚽️ We are launching Thierry Henry World Cup AI Clone, built to cover all 104 matches with Henry’s football judgment and leading models, from @OpenAI and @AnthropicAI to @deepseek_ai, @Kimi_Moonshot, @MiniMax_AI, @XiaomiMiMo and more. For every match, Henry AI Clone will publish its prediction, market view, position logic, and live trading actions on Questflow, powered by different models across the tournament. Not just picks. Not just analysis. Real positions you can actually follow. If Henry AI Clone’s view matches yours, you can copy its position directly on Questflow and ride the World Cup market with it. 104 matches. 10+ models. One AI Clone. Every prediction, position, and review shown publicly. And here is the fun part 🏆 Drop your guess in the comments. What will Henry AI Clone’s final win rate be after all 104 World Cup matches? When the tournament ends, we will pick the closest prediction and send a mystery reward. Explore the Questflow World Cup hub 👇 next.questflow.ai/event/worldcup… #FIFAWorldCup #Prediction

0G × @MiniMax_AI We’re thrilled to partner with MiniMax to bring frontier AI on-chain through verifiable, privacy-preserving compute. MiniMax M3 is one of the hottest models in AI today: #1 open-weight model on Artificial Analysis, topping OpenRouter Trending.