Drama is back on the menu

Google published benchmarks for its upcoming Gemini 4 Argon model showing it competitive with offerings from Anthropic and OpenAI, though Bloomberg reported Google staff are less than impressed with Argon outside testing environments. OpenAI president Greg Brockman is shuttering his personal donations to the Leading the Future political action committee after his initial $25 million gift, following employee discontent that the PAC didn't reflect OpenAI's values. Meta is claiming its data center investments as tax-favorable R&D costs, saving the company billions, after an earlier IRS dispute over counting CEO Mark Zuckerberg's compensation as an R&D expense.

Drama is back on the menu And: Are tech companies cheap? + Shame on Meta Thursday. Welcome to the first day of Q4 Earnings season is just around the corner, and the IPO window is staring down the holiday pause. Let's see who gets out in time. Today, we're talking about the drama consuming Silicon Valley and why old-school venture rules existed , the value of tech companies, and we'll close with a few positive notes on commercializing space. Your friend, Alex. - 📈 Trending Up: OpenAI hacking governments by accident https://transluce.org/us-canada-gov ... Micron's business https://finance.yahoo.com/markets/stocks/article/micron-just-blew-wall-street-away-again--heres-whats-next-091328638.html ... 4x markups https://www.axios.com/pro/enterprise-software-deals/2026/09/30/salesforce-listen-labs-fin ... protecting humans from bots https://techcrunch.com/2026/09/30/reddit-is-killing-rss-feeds-ending-public-api-access-because-of-ai-bots/ ... government borrowing costs https://www.cnbc.com/2026/10/01/us-treasury-bond-yield.html ... performative politicking https://www.cnbc.com/2026/09/30/california-gavin-newsom-ai-ban.html ... fuel prices https://www.cnbc.com/2026/10/01/oil-prices-today-wti-brent.html ... hearing glasses https://techcrunch.com/2026/10/01/hearing-tech-startup-legato-launches-its-ai-hearing-glasses/ ? ... - 📉 Trending Down: Shock https://www.ft.com/content/534b6887-63ac-49c7-a816-a81edd2e8de1 ... rule of law https://abcnews.com/Politics/trump-pardoning-administration-members/story?id=136915372 ... not bombing Iran https://www.bloomberg.com/news/articles/2026-10-01/trump-says-ramped-up-bombing-of-iran-possible-after-midterms ... paying your own way https://www.cnbc.com/2026/10/01/broadcom-lending-anthropic-42-billion-chips-reuters.html ... the free press https://x.com/AaronBlake/status/2105623196901024042 ... Shame on Meta: After trying to argue that its CEO's compensation should count as a tax-favorable R&D expense and a resulting argument https://www.taxnotes.com/research/federal/other-documents/other-court-documents/irs-seeks-summary-judgment-zuckerberg-stock-option-issue/7sk2j with the IRS, Meta is claiming its data center investments are https://www.nytimes.com/2026/09/30/technology/meta-ai-data-centers-taxes.html also R&D costs instead of normal, non-tax-friendly spend https://www.nytimes.com/2026/09/30/technology/meta-ai-data-centers-taxes.html . As a result, Meta is saving billions. - Come the fuck on, Meta. Other tech companies aren't taking as aggressive a tax-prep route, to their credit. I know the IRS has been gutted and the current administration is corporate-friendly, but what the hell. Notes from the AI trenches: Google showed off benchmarks for its upcoming Gemini 4 Argon model https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/ , which appears competitive with offerings from Anthropic and OpenAI after spending months languishing behind the frontier https://www.axios.com/2026/09/30/google-gemini-4 . Google is testing the model with cybersecurity partners, though its own staff is less than impressed with Argon https://www.bloomberg.com/news/articles/2026-09-30/google-grapples-with-employee-skepticism-about-new-gemini-model?accessToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzb3VyY2UiOiJTdWJzY3JpYmVyR2lmdGVkQXJ0aWNsZSIsImlhdCI6MTc5MDgwNTU0NiwiZXhwIjoxNzkxNDEwMzQ2LCJhcnRpY2xlSWQiOiJUTTRNM0lLSzNOWTkwMCIsImJjb25uZWN0SWQiOiJYTENRT1MxQTNQNU9FVkZKUUVTQlNVVU5HVzRPV1NIMiJ9.ZcUfBhNnvuR8AZd8Uo4RO4LePmNKZ7sq7KOfVQYKv6I outside testing environments. - OpenAI's Greg Brockman is shuttering his personal donations to the Leading the Future political action committee https://punchbowl.news/article/tech/open-ai-distances-ltf/ PAC after his initial $25 million gift. Employees had expressed discontent that the PAC didn't reflect OpenAI's values; Brockman said the group had become a " distraction https://www.nytimes.com/2026/09/30/technology/openai-brockman-super-pac-leading-the-future.html ." - Meta CEO Mark Zuckberg and House Speaker Mike Johnson were the seed from which the White House AI plan grew https://www.semafor.com/article/09/30/2026/how-zuckerberg-shaped-trumps-ai-industry-pledge ; other AI CEOs pressed Anthropic's Dario Amodei during the confab about "why he was being so extreme in public about the capabilities of AI models and risks they posed," the WSJ reports https://www.wsj.com/tech/ai/tech-ceos-privately-questioned-amodei-for-sounding-ai-alarm-bells-aaa47df3 , to which Amodei said, "it is important to be honest with the public about model capabilities and not play down risks," the paper writes. - OpenAI accused Moonshot https://openai.com/index/disrupting-a-coordinated-model-distillation-campaign , the Chinese AI lab behind the Kimi family of AI models, of aggressively distilling its own work over the summer. The American company detailed its work to limit other labs' ability to extract intelligence from its models; some AI leaders view such activity as mere competition, others less charitably. Closed models have a clear intelligence edge https://x.com/alex/status/2105391880850567581 over their open rivals at present, perhaps because distillation is a closing door. Silicon Valley is melting down on X I love mess. The often buttoned-up world of startups and venture capital blew its top yesterday after a CEO accused a board observer of deception while firing him. Publicly. The rival company and its investors fired back. Ironically, some of the loudest screeches came from an investor in the original company. Let me explain. Matan Grinberg is the CEO of Factory https://factory.com/ , an AI startup that offers agents for software development. Claiming 100x revenue growth in the last year. Grinberg wrote on X yesterday https://x.com/matanSF/status/2105335179502064038 that he was terminating his board observer, Christopher Degnan. Degnan, known for being the first sales hire at Snowflake, told Factory that he was also talking to Cognition, a Factory AI rival, but that it wasn't a serious commitment. Later, Degnan "called Grinberg to say that the conversation with Cognition that was initially described as casual and one-off was actually formal and recurring." Degnan got the boot from Factory as he announced https://www.linkedin.com/feed/update/urn:li:activity:7511113212096847872/ that he was taking on the Chief Revenue Officer role at Cognition. Now, shuffling executives is not new. Stating publicly that you are firing an advisor over ethical missteps is. Enter the pushback. After Grinberg made his case, Khosla investor and founder Vinod Khosla, whose firm has invested in Cognition https://www.linkedin.com/posts/excited-to-be-investing-in-scott-wu-and-the-share-7189027381599264769-0-IL/ , went nuclear: Grinberg replied that he would "rather not post email receipts, but could assure Vinod that his accusations are factually incorrect." Degnan came to his own defense, of course https://www.linkedin.com/feed/update/urn:li:activity:7511113212096847872/ . Now things get wild. Khosla was also a named investor in Factory's recent $200 million, $5 billion valuation round announced in mid-September https://factory.com/news/5-billion-valuation . Of 2026 . Yes, Khosla's founder is calling a Khosla portfolio company and CEO struggling, unethical, and desperate. Not good Vinod's reputation for being, shall we say, spiky came to the fore with CEOs like Amplitude's Spenser Skates https://x.com/spenserskates/status/2105416200566297064 one of my favorite tech execs saying that the saga was "another great reason not to raise from Vinod," who he described as one "of the biggest bullies of founders in the VC ecosystem." Menlo's Matt Murphy weighed in on Factory's side https://x.com/mmurph/status/2105362626532495788 . Subscribe to keep reading Everything below is for paid subscribers