Everyone from OpenAI to SpaceX is building their own chips | Equity Podcast

OpenAI's Jalapeno chip targets inference costs

OpenAI partnered with Broadcom to develop a custom inference chip named Jalapeno, aiming to lower operational costs and reduce reliance on Nvidia GPUs for specific coding and inference tasks.

Industry hedges against single-supplier risk

Following Apple's vertical integration playbook, companies like SpaceX and Google are designing in-house silicon to customize hardware for specific AI needs and diversify beyond Nvidia.

Cost savings prioritized over consumer benefits

While custom chips aim to reduce infrastructure expenses, savings will likely boost profitability for OpenAI's upcoming IPO rather than lower prices for end users.

Memory shortage drives hardware prices higher

Skyrocketing demand for memory chips has created severe supply constraints, forcing Apple to raise consumer prices due to unavoidable cost pressures from suppliers like Micron.

AI lab costs threaten consumer pricing

With memory constraints affecting the entire chip ecosystem, AI companies may soon pass infrastructure costs to consumers through higher subscription prices for services like ChatGPT.

Groq raises $650M after Nvidia talent deal

Despite losing key executives to Nvidia in a reverse acqui-hire, Groq successfully pivoted to an AI cloud platform and secured significant new funding to lease compute capacity.

SpaceX monetizes compute post-IPO

Following its volatile public debut, SpaceX signed deals to rent out excess compute capacity to smaller AI players, joining the 'neocloud' gold rush to generate immediate revenue.

Allbirds bankruptcy leads to cloud rebirth

The footwear company emerged from bankruptcy restructuring as a neocloud provider, exemplifying the desperate rush to lease compute resources to cash-flush AI companies.

SoftBank questions space-based infrastructure

SoftBank CEO Masayoshi Son publicly challenged the logic of orbital data centers, noting the immense costs and multi-year timelines make them irrelevant to immediate compute shortages.

Starlink synergy drives SpaceX strategy

Critics suggest SpaceX's orbital data center ambitions primarily serve to guarantee launch business for its Starlink satellite replacement cycle rather than solve practical AI infrastructure needs.