✳flâneur — a map of the web's best reading

Computational Power and AI - AI Now Institute

ainowinstitute.org · 14,536 words · saved by 1 readers

Amid a steadily growing push to build AI at larger and larger scale, access to compute—along with data and skilled labor—is a key component2 in building artificial intelligence systems. It is profoundly monopolized at key points in the supply chain by one or a small handful of firms.3 Industry concentration acts as a shaping force in how computational power is manufactured and accessed by tech developers. As we will show, it influences the behavior of even the biggest AI firms as they encounter the effects of compute scarcity. A recent report from Andreessen Horowitz describes compute as “a predominant factor driving the industry today,” noting that companies have spent “more than 80% of their total capital on compute resources.”4 This concentration in compute also incentivizes cloud infrastructure providers to act in ways that will protect their dominant position in the market, racing to release products before they’re ready for widespread use and behaving in ways that encourage lock-

Computational Power and AI - AI Now Institute By Jai Vipra & Sarah Myers West September 27, 2023 In this article What is compute and why does it matter? How is the demand for compute shaping AI development? What kind of hardware is involved? What are the components of compute hardware? What does the supply chain for AI hardware look like? What does the market for data centers look like? How can demand for compute be addressed? How are governments responding? What are the policy implications? Download full PDF report here . What is “compute”, and why does it matter? Computational po

Explore this link on the map →

saved by

related reading