Rory O'Driscoll

Great to see @0xjohnkim on @tbpn today

Youtube 👉 youtu.be/lTuS7Ns8cZg Spotify 👉 open.spotify.com/episode/2Wj7Ei…

@FinnMurphy12 Dublin vcs making snotty comments to budding entrepreneurs from down the country, hundreds of years in.

limerick fintech entrepreneur discovers blockchain technology 18 years into the cycle - says ‘it’s the next big thing’

Congrats to @0xjohnkim & @lijeffrey39 and the entire Paraform team. Recruiting is exactly the kind of knowledge-dense market where AI can collapse coordination costs, but still have humans drive the outcome. Paraform is building the future of recruiting, for every role and industry. Proud to lead their Series B. More from @siddharthvader_ here: scalevp.com/blog/hire-on-e…

We’re excited to lead @paraformtalent's $40M Series B as they scale to solve one of the hardest problems for companies: recruiting. Proud to back @0xjohnkim & @lijeffrey39 youtube.com/watch?v=eXj3dt…

The SaaSpocalypse has left every legacy SaaS company asking the same question. Can you transition existing customers to the new AI way of automating things and save yourself in the process? Wix is a perfect lab experiment for this. They have 6.1 million customers already paying them to build websites, and an AI product in Base44 that has reached $100M ARR doing exactly what those customers now want to do with AI. The use case couldn't be more tailor-made. And the math is straightforward enough. If Base44 can grow like Lovable or Replit, Wix could become a 30% growth company in 24 months. If it can’t, the core will keep declining. This is as clear a second act play as it gets. If you can't cross-sell existing customers to the new AI way of doing the same thing with 6.1 million customers and that obvious a use case, the second act story doesn't work for anyone. There are two trillion dollars worth of public companies and another trillion dollars worth of private companies for whom this is the bet.

@FrancisPSantora Not if the spend is an annual spend ie $600 bn per year. If it is not an annual spend ie if you think the capex is a one off then the nvidia out year estimates are wrong.

@rodriscoll You may want to stretch out the payback period because the chips will be good for around 6 yrs. This requires a much smaller productivity boost per worker.

Three seemingly contradictory thoughts on AI Capex after this week’s discussion on the @twentyminutevc pod: 1. The current level of investment is not sustainable in the long term 2. No one is blinking right now 3. In the long term, the foundation model companies should still be wildly profitable

The land grab is on. Everyone is focused on building more data centers. But Aurelia Han and @siddharthvader_ make the case that the more interesting opportunity is getting more compute out of the ones we already have: scalevp.com/blog/the-big-d…

This does not mean the foundation model companies are not great businesses. See this great post from @jaminball at Clouded Judgement. Long term my guess is Anthropic and OpenAI should be very profitable companies, subsidized by massive over investment from hypescalers desperate to sell them a commodity product that will in fact take away their own business (take a bow, Microsoft). cloudedjudgement.substack.com/p/clouded-judg…

Game theory says keep playing. Only six or seven CEOs matter in this discussion and they've all decided to ante up. Until someone looks at their hand and decides to fold, we're going to overinvest. That's not happening in 2026, so no, the pause in a datacenter in Texas is not a sign of a slowdown.

The only podcast you have to listen to every week. No politics. Just tech. - Anthropic vs The Pentagon: Who Wins - The Data Centre Arms Race: Is the Capex War Stalling - The Era of Public Company Deceleration is Dead - The Ultimate Stock Picks: What to Buy Spotify 👉 open.spotify.com/episode/57WlM4… Youtube 👉 youtu.be/R8YC8OCnDxM?si… Apple Podcasts 👉 podcasts.apple.com/us/podcast/20v… This weeks show with @jasonlk and @rodriscoll 👇

Every VC is understandably scared of making investments that will get crushed by the model companies. So they’re searching for things that feel protected. I wonder if that’s the right approach. Focusing on downside protection has generally been a terrible way to invest in startups.

@Julius_Sten1 Agree, and that is because, unlike most pre-2022 companies, they were AI-forward from day one and attuned from a tech and business model perspective to just add modern AI. Be like Palantir

@rodriscoll $PLTR is literally the only company integrating foundation models at scale with real outcomes. So I suggest option 6 - being a 23 year old company implementing option 3

Intelligence, generated by foundation models like Claude, will infuse all software over the next decade. That’s a given. The question is how it gets to the enterprise. There are five (non-exclusive) paths for this to happen. Enterprises can: 1. Buy directly from the foundation models (the models “just work”) 2. Build it for themselves on top of foundation models (AI for proprietary advantage) 3. Buy from new companies, founded post-2022, built on top of foundation models 4. Buy from existing software vendors who integrate AI into existing apps 5. Buy the outcome from a full-stack, AI-enabled provider (AI lawyer, etc) Options one or two imply the death of all the non-foundation model application software business (“aka software is dead”). Option three implies all pre-2022 application software is dead or at best legacy (“SaaSpocalypse”). Option four says you buy the WCLD index at 8x EBITDA, and option five has you planning AI-enabled roll-ups. Figuring out, market by market, how this plays out and why is where the venture alpha is for the next five years. Hint: answers will vary for reasons that in retrospect will be obvious, and the pace of adoption will differ significantly by market.

@edsim You are absolutely right. The models continue to eat more of the value. As Ilya says, "the models just want to learn". Any investment in a startup pursuing options three to five is an implicit call on defensibility versus the models over the next five plus years. This is the hardest part of every bet right now.

@rodriscoll @marcepntoja 🫡 the timing and speed at which markets dynamically move from bucket to bucket based on each new release is also one of the greatest unknowns and threats in this framework

A few more thoughts on the SaaSpocalypse. The SaaSpocalypse is really the story of a breakup. Wall St is getting ready to fall in love with AI, and to do that, it had to fall out of love with SaaS. Wall St is fickle, but it is serially monogamous. Love is blind; dead love is not. Wall St is suddenly shocked to discover SaaS has stock based comp that should be accounted for in free cash flow, finite TAMs, and companies that mature and grow stale. Meanwhile, Wall St is about to take AI companies public at 50x+ revenues, that dispense $10M stock-based comp packages to employees like lollipops, while having capex bills and negative free cash flow larger than any capex since we built the railways. Why? Because AI is growing like a weed. And growth, like beauty, is, as the poet Keats said, “all you know and all you need to know.” In the face of beauty or three years of 10x growth, all objections fade. We are all in the business of growth.