← Analysis
Flying cars vs. 140 characters
In the same summer of 2011, two venture capitalists published opposite theses about what their own industry was actually funding. Fifteen years of real deal data since then have vindicated pieces of both -- and the industry never actually settled the argument, it just keeps moving the number.

This site has already covered how venture capital became an industry at all -- two 1978/1979 policy decisions that let pension money into the asset class at scale. This is the follow-up question: once that capital was flowing, what did it actually get pointed at?

On July 30, 2011, Peter Thiel's firm Founders Fund published a manifesto with a line that outlived the essay: "We wanted flying cars, instead we got 140 characters."[1] The claim was that venture capital had stopped funding the physical, ambitious, capital-intensive future -- rockets, energy, biology -- and settled for funding "features, widgets, irrelevances."[1] Three weeks later, on August 20, 2011, Marc Andreessen published the opposite argument in the Wall Street Journal: "Why Software Is Eating the World."[2] His case wasn't that hard technology didn't matter -- it was that software was the mechanism by which every other industry, including the physical ones, was about to get restructured. Same summer, same industry, two of its most prominent investors, flatly opposed theses.

The number both men were actually arguing about Neither was speaking from a vacuum. Using PitchBook's own deal data, venture firm MaC Venture Capital found that in the 1990s, 90% of the top 100 venture deals by count -- and 87% by dollar volume -- went to non-physical companies, many of them dot-com startups funded on an idea and a website. Total VC funding went from $8 billion in 1995 to $105 billion in 2000, then the Nasdaq lost 78% of its value by 2002. By the 2000s, that non-physical share had already fallen to 52% of deals and 54% of dollar volume -- a real, if partial, correction, years before either manifesto was written.[3]

So which one turned out to be right? By MaC's count, by the 2000s, 90% of the top 10 most valuable venture-backed companies made software -- Thiel's complaint had real numbers behind it. But looking at 2000 through roughly the last few years as a whole, venture dollars flowed into software companies at about 3x the rate they flowed into hardware -- Andreessen's read that software was where the money and the restructuring both were also had real numbers behind it.[3] The two claims aren't actually contradictory: one is about capital allocation, the other is about which sector's tools ended up rearranging the rest of the economy. Both things happened, at the same time, funded by mostly the same pool of money.

What's changed since is the part neither manifesto could have predicted: deep tech's share of global VC funding was roughly 12% in 2016; by 2026 it had climbed to 36%, per VC firm Celesta's own portfolio research -- nearly a threefold increase in a decade.[4] MaC's count of the top 10 most valuable venture-backed private US companies now includes four hardware-focused ones, up from a low point of essentially zero in the 2000s.[3] Thiel's "flying cars" complaint didn't get answered by anyone changing their mind -- it got answered by capital, over a decade, actually moving.

3x
more VC dollars into software than hardware, 2000-recent (PitchBook / MaC VC)
12% → 36%
deep tech's share of global VC funding, 2016 → 2026 (Celesta VC)
0 → 4
hardware-focused firms in the top 10 most valuable VC-backed US private companies, 2000s → today (MaC VC / PitchBook)
What Oluwadi and Osparna's own graph shows -- and what it can't tell you yet This site's own company graph, built out through Oluwadi and Osparna, isn't a national sample -- it's built region and sector at a time, so it can't stand in for PitchBook's national numbers above. But one pattern in it is worth naming: among funded companies in the graph, Biotech/Pharma has more individually funded companies on record than any single software category -- more than AI Application, more than Enterprise SaaS.[5] We're not citing dollar totals from our own data here, on purpose: the graph currently has duplicate funding-round records for at least one company (a single Anthropic round logged multiple times from separate news mentions), which would overstate any dollar figure we quoted until that gets cleaned up. Company counts aren't affected by that the same way, so that's the number we're standing behind. The gap between "here's what we can say" and "here's what we can't yet" is itself real information, not a footnote.

The honest version of "which industries get support" isn't a single answer -- it's at least three different questions that get collapsed into one. Which sector gets the most deals? Which gets the most dollars? Which sector's few outsized bets produce the founders and companies everyone remembers? Thiel and Andreessen were each answering a different one of those three, in the same summer, without saying so.

Who's on the lever Nobody. That's the actual finding. Neither Thiel's manifesto nor Andreessen's essay moved the capital -- the capital moved because LPs (the pension funds and endowments actually writing the checks into venture funds) allocate toward whatever thesis the last cycle proved out, then take years to notice that the market already moved on. The 2016-2026 swing back toward deep tech wasn't anyone winning the argument; it's the same structural lag already documented in how VC became an industry in the first place -- a multi-year gap between when capital moves and when anyone can tell whether it moved toward the right thing.
Sources
  1. Business Insider, "Founders Fund manifesto coverage", July 30, 2011
  2. Marc Andreessen, Wall Street Journal, August 20, 2011, "Why Software Is Eating the World"
  3. MaC Venture Capital, 2025 (PitchBook-sourced deal/valuation data), "The State of Technology & Culture: A Return to Hard Tech"
  4. Celesta VC, April 2026, "Inside Deep Tech Market: What the Data Tells Us In 2026"
  5. Oluwadi/Osparna company graph, funded-company counts by category — queried directly from this site's own database, 2026-08-09. Category taxonomy re-audited and corrected the same night; dollar totals withheld pending funding-round deduplication.