Thill的下载：软件创新低；看好META、AMZN、GOOGL；TEAM检查喜忧参半

Software hit new lows last week & semis rallied (IGV down 7% vs SMH +11%), w/the AI build-out still in early stages, shown by CRWV deals w/ META & Anthropic.We got positive signals from META w/ new AI model, AMZN w/ bullish CEOletter, & GOOGL w/ strong cloud/ad checks. TEAM partners had a mixed Q. Wewill attend the JEF Private Growth Conf w/ top private cos incl Databricks andCanva user conf in LA. Watch the video below for an interactive download. Muse Spark launch reinforces our thesis that META's AI investments will pay off.Ahead ofthe release, with investor skepticism around META’s model execution especially with scale ofAI investments, it had become clear that Meta Superintelligence Labs was an overhang on thestock. But we were convinced that beyond having a superb ROIC within the core ad business,with leading talent, massive compute, proprietary data, and unmatched distribution, META had theingredients to close the gap in AI. Our initial feedback is that Muse Spark is a better-than-expectedleap in performance in catching up with SOTA models on many fronts, and shows an impressivetrajectory for future releases. While we note it lags on coding and longer-horizon agentic tasks,its improvement in reasoning with token efficiency & multimodal strength makes it well suited forscaled deployment. TEAM: friction in the partner ecosystem, but could be a net positive for the co to take more dealsdirectly.Key takes:1)Partner sentiment negative this Q as recent RIF gutted the partner team &Oct. comp changes reduced % margin partners have on deals. While this may cause friction in thepartner ecosystem, we think it could be a net positive for TEAM to take more deals directly.2)C1Qresults mixed: one partner saw their best March ever as geopolitical tension caused a surge in spendfrom defense contractors; the other saw their worst Q in years as TEAM took several deals directly.3)Cloud migration tailwind should last 2-3 years, but partners expect growth to stagnate beyondthat if TEAM doesn't create another product for them to sell.4)Both partners believe TEAM needsto move to consumption pricing.More inside. Positive cloud check: GCP better positioned for acceleration; GPU scarcity pushes multi-cloudadoption.Key takes:1)GCP is better positioned for faster growth vs. AWS & Azure, as its TPU-backed infra enables Vertex to deliver more Opus tokens at scale, making GOOGL the largestprovider of Claude & overall tokens among the expert's customers.2)GPU scarcity & higher pricingare forcing production workloads into multi-cloud setups, w/ custs increasingly leveraging neo-clouds & piping data back to hyperscalers.3)Supply constraints may ease in Q3/Q4 w/ the rolloutof Rubin, though stronger-than-expected AI deployments & spend across ISVs, digital natives, &enterprises could prolong pressure.4)SNOW Intelligence is underappreciated w/ the potential todrive a meaningful demand ramp across non-technical users.More inside. CRWV +26% last wk as AI demand continues to outstrip supply.The co. expanded its METAdeal by $21B, bringing total committed spend by META to $35B thru 2032, and signed a multi-yragreement w/ Anthropic to support Claude development/deployment. We estimate total backlognow exceeds $90B vs. $67B last qtr. CRWV also raised an upsized $3.5B convertible note. See inside for key takes from TEAM, Cloud, & Ad Expert calls, Cheeky Pint Podcast w/ GOOGL CEO, &our thoughts on ORCL CFO change. See here for key takes on this year's HumanX AI Conference. We hope you can attend our AI + Software + Internet Conference May 26-28 in Newport Beach! Email TEAM Recap We spoke to 2 TEAM partners this week, one boutique and one large, for mixed checks. See speakerlinks here and key takes below: 1. Partner sentiment negative this Q due to adverse impacts from RIF and compensation changes.Many of the people that partners worked with from TEAM's partner program were let go as part of the10% RIF announced last month. This, coupled with October compensation changes that reduced theamount of margin customers have to work with from 10-25% to 5%, is causing partners to feel likeless of a priority at TEAM. Despite the friction in the ecosystem, we think TEAM stands to benefit fromtaking more deals directly. 2. Partners see mixed C1Q results.One partner saw one of their best C1Qs while the other saw one oftheir worst. The boutique partner saw C1Q come in substantially ahead of expectations as geopoliticaltensions resulted in an unlimited budget for defense customers. Keeping in mind that 50% of his baseis public sector, the partner noted that the rest of his base performed as expected. The larger partnercame in below expectations as TEAM took multiple large deals directly. 3. Questions remain over what LT growth will look like if partners don't have another product tosell.Partners were mixed when asked if TEAM can sustain 20%+ growth, but both agree that growthmay start to taper off as Cloud migration tailwinds subside in 2-3 years if