“Meta spent almost as much as the Manhattan Project on GPUs in today's dollars”
Meta’s reported $30B spend on Nvidia GPUs is being compared to the inflation-adjusted cost of the Manhattan Project and Apollo, raising questions about how today’s private tech investments stack up against historic, state-led “moonshots.” Commenters debate whether modern GPUs and EUV lithography rival or exceed the scientific and logistical complexity of building nuclear weapons or sending humans to the Moon, and whether inflation metrics or GDP share are the right way to compare eras. The thread also probes who benefits from such vast outlays—shareholders, users, or society at large—especially when the outputs are ad-targeting systems and large language models rather than public infrastructure or fundamental science.
Scale and Cost Comparisons
- Meta’s ~$30B GPU spend is compared to Manhattan Project and Apollo, but several note:
- US GDP is ~9× higher than in 1945, so similar nominal amounts are relatively “smaller”.
- Manhattan and Apollo each peaked around 0.4% of GDP; Meta’s outlay is ~0.1% of current US GDP and spread over years.
- Others compare against:
- Apple’s $110B stock buyback, making Meta’s spend less shocking.
- IBM System/360 development (said to exceed Manhattan in 1960s dollars).
- Some argue such historical cost comparisons are misleading without context on what money, wages, and tech could buy at the time.
Technological Complexity: GPUs vs Nuclear & Space
- Several argue modern GPUs, EUV lithography, and fabs (TSMC, ASML) are as impressive or more complex than Manhattan-era nuclear work.
- Counterpoint: Manhattan’s real scale was in industrialization (Hanford, Oak Ridge) and novel uranium/plutonium production, plus long‑term cleanup costs.
Economic & Inflation Debates
- Big subthread on whether official inflation understates past price growth:
- One camp uses median family income, gold, housing, education, healthcare to claim heavy undercounting.
- Another defends CPI and modern economic methods; stresses productivity gains, better goods, and rising living standards.
- Disagreement on whether today’s higher project prices reflect true inflation vs higher real wages, safety, regulation, and richer societies.
Government vs Corporate Mega‑Projects
- Wartime projects seen as uniquely focused, with little tolerance for waste or delay; today’s peacetime bureaucracy, lobbying, and regulation are blamed for higher costs and slower progress.
- Others note Manhattan and Apollo were also wasteful and pushed many costs (e.g., environmental cleanup) into the future.
- Some worry that large corporations now rival or exceed states in economic power and can pursue “moonshots” driven by ad revenue rather than public goals.
Meta’s AI Strategy and Impact
- GPUs fund recommendation systems (Feed, Reels), not just chatbots; some frame this as “weapon‑grade” optimization of user addiction, especially for kids.
- Mixed views on Meta open‑sourcing LLaMA:
- Praise: massive private spend yielding public AI models; more efficient than closed efforts.
- Concern: easier mass sockpuppeting, manipulation, and uncertain societal value.
- Debate over whether Meta is “behind” OpenAI or simply optimizing for different products and revenue streams.
Social, Ethical, and Environmental Concerns
- Skeptics see billions in compute and energy spent on “shitty chatbots” and attention‑hacking rather than housing, health, or climate.
- Others argue risky private bets are exactly how technological progress happens, even if many feel current priorities squander human potential.