# Silicon Valley discovers peer pressure > Source: > Published: 2026-07-27 17:19:00+00:00 # Silicon Valley discovers peer pressure ## Today, we're talking Nvidia, livestreaming, Shein's IPO, and the open-weight AI mess. **Monday.** We’re back in another * talks* phase of the U.S.-Iran war. Stocks are up, oil prices are down. Let’s see if this round of detente goes anywhere. Back on our turf, it’s a [huge earnings week](https://www.nasdaq.com/market-activity/earnings). Today, **F5** reports. Tuesday brings **Visa**, **Boeing**, **Corning**, and **PayPal **(the Stripe target). Wednesday will see **Microsoft**, **Meta**, **Arm**, **Qualcomm**, **Fortinet**, **SoFi**, **Procore**, **Lemonade**, **Tenable**, **Sonos**, and **Coursera** reporting. And Thursday will have **Apple**, **Amazon**, **Mastercard**, **Coinbase**, **Roblox**, **Reddit**, **Strategy**, **Rivian**, **Blue Owl**, **GoDaddy**, and **Riot Platforms**. Today, we’re talking Nvidia, livestreaming, Shein’s IPO, and the open-weight AI mess. To work! **— Alex** ## 📈 Trending Up [Good summaries of the OpenAI-Hugging Face hack](https://m-mitchell.com/HF-hack-cartoon/)…[paying to clean up your own mess](https://x.com/ClementDelangue/status/2081056675558195657)? …[new education heads in India](https://www.cnn.com/2026/07/25/india/india-education-minister-resignation-intl)…[holding hands](https://www.reuters.com/business/nvidia-forms-industry-alliance-open-ai-security-after-hugging-face-hack-2026-07-27/)…[corporate hiring](https://www.wsj.com/business/big-companies-are-starting-to-hire-again-defying-predictions-of-ai-wipeout-f4974e99?mod=hp_lead_pos3)…[Anthropic’s gross margins](https://x.com/i/status/2081454984521282012)? … **Nvidia customer support:** With Microsoft trying to replace third-party AI models in its products with its own models, OpenAI is turning to other technology giants to help finance its growth. Enter Nvidia, once again the world’s most valuable company, with a reported [$250 billion financial backstop](https://www.wsj.com/tech/ai/nvidia-in-talks-with-openai-to-guarantee-250-billion-financing-for-data-center-3dd6eae3?st=pTU9bH) for a 10-gigawatt data center complex in Ohio. The idea is simple: SoftBank unit SB Energy is building an [enormous data center for OpenAI](https://portscampus.com/), which the AI lab has agreed to lease for years. SB Energy wants to borrow against pledged lease payments, but what if OpenAI can’t pay? That’s where Nvidia has stepped in with its commitment to backstop OpenAI’s payments. Nvidia stands to generate more long-term cash flow if AI thrives, and one way to shove the technology forward is to help reduce its cost of use. Lower token prices, in other words, which more compute capacity will help with. If you believe in the oft-cited Jevons Paradox (improved usage efficiency lowers the cost of consuming a resource, raising demand and potentially driving greater total consumption), getting the cost of AI inference by swelling compute capacity may make a good way to ensure token demand. - In this case, we’re assuming economies of scale at play given this is a multi-gigawatt data center, and that new Nvidia chips and related gear will offer better economics than preceding generations. The risk is that Nvidia is sticking its neck out for a partner (customer) that has a history of over-committing and later retrenching its compute commitments (See: [The U.K.](https://www.theguardian.com/technology/2026/jul/04/openai-apparent-failure-visit-key-site-questions-stargate-uk-project)). On the other hand, Nvidia wants to sell a lot of chips, and if it can help its customers undergird their borrowing, why not? The *why not* in this case is the risk that the AI boom slows, and the chip giant is stuck [holding the bag on behalf of all its myriad customers](https://www.datacenterdynamics.com/en/news/nvidia-acts-as-backstop-for-customer-gpus-in-return-for-cut-of-cloud-revenue/). - Speaking of Nvidia making bets, the company is [investing in Safe Superintelligence](https://www.wsj.com/tech/ai/nvidia-bets-on-ilya-sutskevers-new-ai-lab-to-expand-compute-reach-f95596e8), an OpenAI-spinoff that has yet to release much of anything. **Quasi-journalism:** In the time between TBPN’s surge to the top of the charts and the launch of MTS, a European technology-friendly and geographically partisan media group called ‘[etn](https://x.com/etnshow)‘ was launched. Unsurprisingly, the group focuses on European startups, and it [just raised $1.6 million](https://x.com/etnshow/status/2081681428450808294) in a seed round to build “Europe’s first daily live technology show.” New facilities are coming online in London soon. On one hand, that’s great to see. Many American technologists and investors regard the European technology scene as several cheese-producing goats with an intern, so a bit of continental chest-thumping doesn’t bother them. But it isn’t great to see booster-style livestreaming fill the space that should be stuffed with actual journalism. Etn is necessary, but insufficient. Still, a group like etn can at least raise money today. I doubt a publication that wanted to cover the same companies could manage that. Would angels [from OpenAI and Deepmind](https://techcrunch.com/2026/07/27/europe-got-its-own-tbpn-style-live-show-and-its-already-a-hot-spot-on-a-press-tour/) show up for the latter? **Chip companies?** The South Korean stock market is [gyrating](https://www.koreajoongangdaily.com/business/kospi-closes-up-nearly-1-on-tech-buyup-following-san-francisco-summit/12793625) while investors try to affix a reasonable price on memory companies with famously boom-bust economics. Meanwhile, Chinese DRAM producer ChangXin Memory Technologies went public on the [tech-heavy Shanghai STAR board](https://www.bbc.com/news/articles/c9q9w3x9qn2o). Its shares rose nearly 500% in their first day, making the company “the most valuable China-listed company,” [per CNBC](https://www.cnbc.com/2026/07/27/cxmt-china-market-debut-chipmaker-ipo.html). - CXMT said it had 7.8% of the global DRAM market in Q4 2025; given the company’s ~$487 billion market cap, we can infer a $6.2 trillion global memory market cap. Which, ahem, seems a bit elevated. - Meanwhile, while Chinese investors snapped up shares in ChangXin, [semiconductor stocks](https://finance.yahoo.com/quote/SMH/)are taking a beating on American markets today, off nearly 3%. [📉](https://finance.yahoo.com/news/servicenow-pledges-1-5bn-investment-110000403.html) Trending Down [📉](https://finance.yahoo.com/news/servicenow-pledges-1-5bn-investment-110000403.html) [Borrowing costs](https://www.cnbc.com/2026/07/27/treasury-yields-oil-iran.html)…[Ukraine-Iran relations](https://www.aljazeera.com/news/2026/7/27/iran-warns-ukraine-of-retaliation-after-deadly-caspian-sea-strike)…[Dutch monopolies](https://www.theinformation.com/articles/china-starts-mass-producing-homegrown-duv-chipmaking-tools-advance-local-chip-industry)…[AI costs](https://pathfounders.com/p/multiverse-computing-lands-570m-series-c)? …[not AI costs](https://x.com/ArtificialAnlys/status/2080734447717298483)…[media consolidation](http://www.cnn.com/2026/07/24/media/paramount-warner-bros-discovery-merger-delay)? …[personal privacy](https://www.cnn.com/2026/07/27/health/emergency-room-records-cpsc)…[Russian GDP growth](https://www.euronews.com/2026/07/25/russian-central-bank-cuts-gdp-growth-forecast-to-zero-expects-faster-inflation)… **Shein-o-nomics:** Fast fashion giant Shein [filed to list in Hong Kong](https://www1.hkexnews.hk/app/sehk/2026/108766/documents/sehk26072600126.pdf?utm_source=chatgpt.com), giving us a peek at its economics. Here’s the income statement: This is Shein *after *it blew up worldwide, after the United States changed its rules regarding low-value imports — [the so-called de minimis exemption](https://www.federalregister.gov/documents/2026/06/24/2026-12670/indefinite-suspension-of-the-de-minimis-exemption-for-merchandise-arriving-through-all-modes-other). You’d want its 2020-2023 numbers to get a handle on its come-up. With that context in mind, we can see how the fast-fashion company’s growth has slowed from 2024 to 2025. Its product revenues have actually *declined* between Q1 2025 and Q1 2026. The company’s profitability is also under pressure, though in operating profit terms, Shein was still in the black in the first quarter, but certain non-cash charges pushed its net profit into the red. Shein is a large business, but it’s going to be tricky to value. At best, its revenue growth is modest, profitability is minute, and trade channels are growing constricted. That’s before we discuss rising materials and energy costs due to global conflicts. Shein will go public, but I’d struggle to tell you what it’s worth. Sans a new growth vector, what would you judge its price tag to be? ## Everyone’s wrong about the open-weight mess The letter dozens of American technology companies signed last week imploring the United States government to be careful with its approach to regulating open-weight AI models (low-cost Chinese AI models) attracted more endorsers after it went live. Here’s [the first list](https://x.com/JensenHuang/status/2080643682408321103): And the most recent [per Microsoft](https://www.microsoft.com/en-us/corporate-responsibility/topics/open-weight/): The new list features American giants, open-source projects, accelerators, unicorns, European companies, public companies, cybersecurity companies, and even inference providers. It’s nearly *everyone* of importance in the technology world. Which makes the missing names all the more interesting. Two big ones stand out for their absence: Anthropic and one of its major backers, Amazon. Microsoft and Alphabet (Google) have each invested in Anthropic, but Amazon remains its core partner (much like OpenAI and Microsoft). On one hand, nearly every technology company is raising its hand in favor of keeping open-weight AI unencumbered — effectively no federal punishment for Chinese companies that are allegedly mass-distilling closed-source American AI models. (If such an effort is even technically possible, let alone economically useful.) On the other hand, Anthropic and a major investor are standing on the other side of the fence. Or as one [OpenAI staffer put it](https://x.com/tszzl/status/2081174295389290630): i don’t think every corporation in capitalism banding together meekly signing petitions complaining against closed models realizes how cool they’re making AI look. unlimited aura farming once again Roon is pointing out that while most tech shops line up with party thinking on The New Current Thing, Anthropic is taking an orthogonal approach. Which makes it appear, in a sense, pretty damn cool. - Calls for Anthropic to ‘ [read the room](https://x.com/DavidSacks/status/2081033408273395858)‘ by David Sacks (former White House AI lead) fell flat after tech folks argued that they had “[fought for years against ‘read the room[.]’](https://x.com/micsolana/status/2081104296159187325)” - This led to [an all-time tweet from Sacks](https://x.com/DavidSacks/status/2081127454907965885). ** CO** has written about how weird ( [neutral-positive](https://www.cautiousoptimism.news/the-headlines-are-wrong-about-anthropics-pitch-to-slow-ai-progress/)) Anthropic is regarding its AI models: It keeps saying they should be considered as nearly [conscious-adjacent](https://www.cautiousoptimism.news/being-public-is-great-actually/), but it’s odd that the company was the first to not release its best models to the public generally, out of sync in its supposed woke-ness. Should we be surprised that Anthropic is off-piste here? No, and one would be wise to recognize the company’s odd bets have paid off. While OpenAI was pursuing what now appear to be [side quests](https://techcrunch.com/2026/07/09/openai-is-shutting-down-atlas-but-its-ai-browser-ambitions-are-still-growing/), Anthropic conquered the agentic coding market. Its rivals are still playing catch-up. Hell, Anthropic [cooked up MCP](https://www.anthropic.com/news/model-context-protocol), now the industry standard for connecting AI to information. If being different (weird?) has worked thus far, why wouldn’t Anthropic stick to its convictions? To wit, the company writes at the top of its homepage that it is building “AI research and products that put safety at the frontier.” Research and safety get two-thirds of the billing! There’s some pretty notable mixed messaging going around. Not only are several companies on the signatory rap sheet big purveyors of closed-source AI, many are jealous protectors of their own IP (and might be a bit more critical of distillation if it happened to *them*). And then there’s OpenAI, which famously moved away from releasing open AI models to a closed-source approach that has [seen it make complaints to Congress](https://assets.bwbx.io/documents/users/iqjWHBFdfxIU/rRmql_jJcxb4/v0) about distillation: It feels a bit off to see OpenAI hopping aboard the Cool Kids Wagon now. But there’s some reasoning behind its signature. The AI lab’s views on distillation hew towards the technical, [per President Greg Brockman](https://fortune.com/2026/07/25/openai-president-greg-brockman-interview-challenges-pressure-high-stakes-ai-startup-hugging-face-regulation/): Now, to the question of distillation: There’s the question of: Is it a technical thing or a policy thing that we should consider it? My answer is that there’s absolutely unquestionably a very core technical component here. We’ve always had distillation protections on our platform within our overall framework of how we secure these models. Distillation is a very nuanced thing. Getting a model that has the exact same capabilities as these core frontier models requires a model that’s exactly the same size compute as one of these frontier models. It’s actually very difficult to do, so we need to recognize the technical facts when thinking about policy. I don’t have a full judgement on what the right answer is, but I think it’s clear as an industry, we need to continuously uplevel our technical safeguards. In short: Yes, China is trying to yoink our work, but it’s on us to protect it. That’s a very measured response, one that I would hazard reflects OpenAI’s desire to avoid being branded a pariah by the American technology industry. Though that’s not something Anthropic seems to worry about very much. All of this is *good*. It’s great to see American tech shops band together to try and influence policy that protects the openness of a market. It’s good to see Anthropic and Amazon take the opposite approach, as the conflict between the two positions will yield more durable policy and products than a single perspective could on its own. Call it an echo of our view that *more competition is more good* — this time seen through the prism of the marketplace of AI regulatory ideas. Who’s right? I think there’s a middle ground to be found here. Onerous limitations on use of open-weight AI models from China would be self-defeating, as you can’t put open software in a locked box. At the same time, Washington could make a little hay out of distillation during trade talks with China. Perhaps it could pressure the labs in question to *chill* a little. Regardless, signing the letter is Valley virtue signaling par excellence. Even if you agree with the policy view in question, admit it for what it is.