Pundit
David Levy
15 receipts • Tracked since Apr 2026
David Levy is founder of Porch Capital and former AWS and Stripe executive with decades of experience in venture capital, startup operations, and AI infrastructure. He is recognized as an expert on enterprise AI adoption and how startup technology choices forecast future infrastructure trends.
Affiliations
No firm affiliations on record.
— Filed —
15 receipts
2026
9 predictions- Jan 7, 2026Story still evolving.“The riskier call: enterprises will hit an adoption tipping point. Why? Because the ROI is too hard to ignore.”
Filed Jan 7, 2026. Levy's "risky" 2026 enterprise-adoption tipping-point call. Will resolve against year-end enterprise AI penetration surveys (Gartner, BCG, McKinsey) and AI-related software ARR growth.
- Jan 7, 2026Story still evolving.“If 2025 was the Year of the GPU Cloud IPO(s), then 2026 will be the Year of Agentic AI… agentic systems don't just generate more tokens; they generate orders of magnitude more inference events.”
Filed Jan 7, 2026 as Levy's headline 2026 call. Agentic adoption and inference-event growth are tracking up but not yet at the "orders of magnitude" step-change he predicts. Re-grade by EOY 2026 against agent-platform deployment data and inference-spend disclosures.
- Apr 30, 2026Too early to tell.“The question wasn't whether OpenAI bought too much compute; it was whether they can generate enough revenue to outrun what they've already committed to. Answer: yep.”
Filed Apr 30, 2026. Levy argues OpenAI's ~$123B/yr compute commitment can be outrun by exploding workflow volume and rising monetization mix. Outcome will hinge on 2026–2028 ARR vs. capex realization — track against OpenAI revenue disclosures and capex actuals.
- Jan 7, 2026Story still evolving.“In 2026, that pressure will push inference workloads outward into CDNs, edge networks, purpose-built inference platforms, on-prem environments, and on-device.”
Filed Jan 7, 2026. Track against measurable shifts in inference deployment topology — Cloudflare/Fastly/Akamai inference revenue, Groq/Cerebras/SambaNova deal flow, and on-device model launches over 2026.
- Apr 28, 2026Too early to tell.“I think the questions being asked now are not like how do you add AI to um Salesforce or how do you add AI to um uh Excel or G but what like what are the native like what I don't even know that like what does a native AI um you know analysis look like? What does a native AI you know like how how does this all look you know if we don't start with the tools we use now?”
Filed for tracking. Outcome to be evaluated as the timeline plays out.
- Apr 28, 2026Too early to tell.“I think like that's where these either, you know, agentic or, you know, rapper type platforms need to be. They need to be in a place where they're going to, you know, unlock more value than using a single platform.”
Filed for tracking. Outcome to be evaluated as the timeline plays out.
- Apr 28, 2026Too early to tell.“I think like as we become more accustomed and more okay with um using those for our own use, you know, I think that'll drop the barrier to when you get a call that's AI as long as it's not, you know, spam spam is spam. Um but like um I think we'd get more comfortable engaging with voice AI as we use it more on our own.”
Filed for tracking. Outcome to be evaluated as the timeline plays out.
- Apr 28, 2026Too early to tell.“I think the tools that people are using today, they're likely choosing tools first that they wouldn't choose before. Uh um and yeah, you know, I'd love to hear what what you guys are seeing. Um you know, I in general categories where it's kind of stupidly obvious. customer service is the first one that comes to mind or maybe the second one. The first one is cloud code but that's less to do with marketing. Uh but uh anything on customer service, anything on sales autom uh automation um that's what I'm seeing like first”
Filed for tracking. Outcome to be evaluated as the timeline plays out.
- Apr 28, 2026Too early to tell.“you can tell what enterprises are going to use um two, three, four, five years from now by looking at startups.”
Filed for tracking. Outcome to be evaluated as the timeline plays out.
2025
1 prediction- Jan 6, 2025Called it.“Welcome to the Year of the GPU Cloud IPO(s) — predicting CoreWeave's IPO and imminent valuation explosion ushering in an AI-related IPO boom.”
Called Jan 6, 2025. CoreWeave IPO'd in March 2025 and ~doubled to ~$80 by year-end; NBIS tripled; APLD/BTBT pivoted to GPU cloud branding; Lambda and Crusoe lined up bankers. Levy's own 2026 retrospective documents the chain of confirmations.
2013
1 prediction- Apr 16, 2013Partially right.“I'm saying this for the third time. Facebook will do search. And [more importantly] they'll do it well.”
Called Apr 16, 2013. Graph Search itself flopped and was effectively retired by 2019, but the underlying claim — that Facebook would build serious search/discovery products — was vindicated by Reels search, AI-assisted feed ranking, and Meta AI's in-app search now serving billions of queries. Right thesis, wrong product vehicle.
2012
1 prediction- Jul 19, 2012Called it.“Is it conceivable that Facebook knew/knows that the ~$4b desktop ad business was not the future of the company and began to focus its resources on sponsored stories/mobile ads that do show up in mobile feeds and can therefore be monetized (and apparently more effectively than their desktop counterparts)? You bet it is… implying that the company has no business model is extremely shortsighted.”
Filed Jul 19, 2012 — two months after the FB IPO flop, with consensus calling Facebook structurally unmonetizable on mobile. Mobile went from ~14% of FB ad revenue in Q2 2012 to ~49% one year later and >80% by 2015; FB stock 4x'd by 2015. The "no mobile business model" thesis was definitively wrong, exactly as Levy argued.
2011
3 predictions- Jul 5, 2011Called it.“Forrester wrote a report in June 2011 forecasting that today's 50 million or so "personal cloud" users will grow to almost 200 million by 2016 driving $12 billion in annual revenue by then… So whatever number you come up with for the personal cloud stuff — double it.”
Called Jul 5, 2011. By 2016 Dropbox alone had 500M+ registered users and Google Drive 800M+; iCloud, OneDrive, and Google Photos each crossed the hundreds of millions. Personal-cloud user counts blew well past Forrester's 200M estimate, validating Levy's "double it" call.
- Jul 5, 2011Called it.“Soon everyone will have a smart phone with a data plan… 80% of the 70% that don't have data plans get plans. That means over 130 million new data plans — at $30/month that's almost $50 billion in incremental revenue per year to the carriers.”
Called Jul 5, 2011 with US smartphone penetration ~30%. Pew put US smartphone ownership at 81% by 2019 and ~90% by 2024; effectively all activated US lines now carry data plans, and US wireless service revenue rose roughly $40–50B over the decade — squarely in the order of magnitude Levy modeled.
- Jul 5, 2011Called it.“The NASDAQ is going to 5K again as investors bid up valuations of companies participating in this expansion and financials cash in on the boom.”
Called Jul 5, 2011 (NASDAQ ~2,800). The index reclaimed its dot-com peak and crossed 5,000 on March 2, 2015 — almost exactly the 5-year cycle Levy described — driven by the social/mobile/cloud names he flagged.