The common thread is **capital chasing strategic bottlenecks while policy risk moves closer to the operating model**. AI needs power and funding, platforms are buying distribution, governments are intervening in energy costs, and U.S.–China tech ties remain selectively open rather than normalized.
Huawei and Qualcomm’s new multi-year patent cross-license reduces some IP friction across 5G, AI and networking. But it is not a broader thaw: export controls remain intact, and regulatory approval still matters.
Uber’s $2.3B acquisition of ezCater is a different kind of infrastructure play—distribution. More than $2.5B in catering bookings and 140K+ restaurant partners give Uber Eats and Uber for Business a higher-ticket corporate channel. The question is how much cross-selling actually converts into margin.
Power remains one of AI’s hardest constraints. Google locked in a 20-year nuclear agreement with Constellation, including 890 MW of planned uprates plus a separate 2.7 GW supply deal. The catch: much of the incremental capacity does not arrive until 2028–32, while data-center demand is scaling now.
Washington is also trying to manage the energy shock directly. Trump temporarily opened highway use of red-dyed diesel and deferred the 24.4¢/gal federal tax through year-end. That helps at the margin, but it does not create more diesel.
Private AI valuations remain aggressive. Moonshot AI reportedly finished a round near a $50B valuation and is considering a ~$5B Hong Kong IPO, while DeepSeek is raising billions of its own. Revenue delivery now has to catch up with compute spending and private-market expectations.
And France’s widening school protests are another reminder that fiscal and social constraints matter too: more than 400 schools faced disruption as pressure builds over staffing, facilities and workloads.
**Capital is abundant for strategic assets. The harder question is whether infrastructure, policy and execution can keep up.**