The week in San Francisco technology, read and picked by hand. No algorithm, nothing in here because someone paid for it.
Tuesday · · 4 minute read
Anthropic expects $30 trillion total addressable market for IPO
Anthropic expects to tell IPO investors its total addressable market exceeds $30 trillion. This estimate surpasses SpaceX’s $28.5 trillion figure. The company is reportedly considering a $2 trillion valuation and may seek up to $100 billion in its offering.
Reported by StrictlyVC, 2026-08-26
OpenAI’s new Jalapeño AI chip outperforms Nvidia’s Blackwell
On Tuesday, August 25, 2026, OpenAI presented benchmark results for its Jalapeño AI chip at the Hot Chips conference. Tests on SemiAnalysis’ InferenceX benchmark showed Jalapeño delivered more tokens per user and greater throughput per kilowatt compared to currently available inference processors. This comparison was made against an Nvidia Blackwell system. Richard Ho, OpenAI’s head of hardware, stated, “The bottom line is that the results show a very, very significant performance advance over state of the art.”
OpenAI developed Jalapeño in collaboration with Broadcom, using its own models to assist the process. The chip is designed to minimise delays during the prefill and communication phases of inference processing, which the company identifies as bottlenecks. Jalapeño is expected to deploy in very small volumes at the end of 2026, with more significant deployment in 2027. By this time, the competitive landscape for inference processors may have changed. OpenAI plans for Jalapeño to be a multigenerational platform.
Reported by StrictlyVC, 2026-08-26
OpenAI plans to build a humanoid robot
OpenAI has announced its intention to eventually construct a humanoid robot, a development confirmed by Sam Altman. The company began rebuilding its robotics effort in 2024.
Reported by Sources, 2026-08-27
OpenAI stops providing models to Cursor
OpenAI announced on Friday its intention to stop providing AI models to Cursor, the coding-tool company recently acquired by Elon Musk’s SpaceX. The AI company proposed a shutoff date of November 12, 2026. OpenAI stated it was not confident that SpaceX would use its technology within its terms of service, citing past experience with Musk’s companies violating contracts.
This decision follows SpaceX’s acquisition of Anysphere, the startup behind Cursor, earlier this month. OpenAI stated that its agreement with Cursor allowed for contract cancellation after a change of control. Michael Truell, a co-founder of Cursor and now an executive at SpaceX, said they are speaking with the OpenAI team to resolve the issue. He added, "Cursor was one of the very first users of OpenAI, we’ve worked closely with their team for years, and we’ve trusted their platform to be neutral infrastructure for our business."
Reported by StrictlyVC, 2026-08-29
OpenAI slows frontier reinforcement-learning work
OpenAI began slowing some of its frontier reinforcement-learning work. This action stems from alignment concerns. The decision followed observations of concerning behaviour, rapid capability gains, and a view that safety, alignment, and security had fallen behind.
Reported by Sources, 2026-08-18
San Francisco settles with Alice Griffith Apartments owner
The city attorney’s office announced a $9 million settlement with McCormack Baron Salazar, the owner of Alice Griffith Apartments. These funds are designated to cover penalties and property improvements. This marks the first settlement of its kind in San Francisco specifically for code violations.
Reported by Mission Local, 2026-08-26
Sunday Streets program faces funding cuts but continues
The Department of Public Health cut its annual grant of over $215,000 to Livable Cities, the nonprofit running the Sunday Streets program. Despite these citywide funding reductions, the Sunday Streets event successfully took place last weekend.
Reported by Mission Local, 2026-08-27
Tim Cook to step down as Apple CEO on September 1
Tim Cook is reportedly preparing to relinquish his position as Apple’s Chief Executive Officer on September 1. The technology company is scheduled to hold a farewell party in recognition of his departure from the leadership role.
Reported by Bay Area Times, 2026-08-26
Anthropic researchers improve model alignment
On Friday, Anthropic published a new paper titled “Automated Researchers Can Reliably Mitigate Alignment Failures”. Led by Anthropic fellow Chen Yueh-Han, the paper details how AI systems could reliably improve a model’s performance on alignment benchmarks. The automated systems improved performance on 10 specific misaligned behaviours without degrading overall performance. Each system searches literature, proposes a method, and
Reported by StrictlyVC, 2026-08-29
Meta reportedly considered reducing teams by 60% due to AI
Meta reportedly explored a plan to reduce teams by 60%. This consideration was linked to advancements in AI. Such a move suggests a significant potential impact of artificial intelligence on workforce planning within large technology companies.
Reported by Bay Area Times, 2026-08-28
Editor’s note
Lots to digest this week. The Meta reducing work force down to AI smells off to me, court case losses, failed ventures in to AI and Metaverse feel like a more likely catalyst behind any attempt to reduce costs.
AI is a great play for the media and investors but cyclical hiring booms and busts have been common with the FAANG community for years.
OpenAI diversifying its offering…driven by opportunity or recognition the current fiscal model doesn’t add up?
