# Tech forgot to ask permission > Source: > Published: 2026-06-18 16:32:13+00:00 # Tech forgot to ask permission * Welcome to *. [Cautious Optimism](https://www.cautiousoptimism.news/), a newsletter on tech, business, and power. Modestly upbeat **Thursday**. Tired? No rest! There’s too much to talk about. Today, we’re diving into Intel’s latest (and totally free-market) win, what companies are doing to help plug the Mythos-sized hole in modern cybersecurity, Microsoft considering Chinese AI models, and why the *global *backlash to data centers is reaching a boiling point requiring new (better?) leadership from tech companies. To work! — Alex ## 📈 Trending Up [Europe’s trade deficit with China](https://www.theguardian.com/business/2026/jun/15/eu-trade-deficit-with-china-record-1bn-a-day)…[bracing for “China shock”](https://www.axios.com/2026/06/16/china-shock-g7-summit-eu)…[fighting over nickel](https://asiatimes.com/2026/06/indonesias-nickel-nationalism-falters-in-the-face-of-chinas-tech/)…[Cuban capitalism](https://www.bloomberg.com/news/articles/2026-06-18/cuba-makes-market-friendly-changes-in-the-face-of-trump-pressure)…[state capitalism](https://www.cnbc.com/2026/06/18/trump-intel-apple-chip-design-deal.html)…[the Strait of Hormuz](https://www.cnbc.com/2026/06/18/iran-deal-strait-hormuz-oil-tanker-saudi-arabia-traffic.html)……*interest rates in Japan* **Intel shares:** The American state chip company Intel announced through its part owner (the White House) that, [to quote POTUS](https://truthsocial.com/@realDonaldTrump/posts/116769225357410422), “Apple has agreed to work with Intel to design and build its Chips in America.” Neither company has confirmed the news, implying that the details may be a little unvarnished. What matters more for our purposes is that POTUS crowed in his post on his personal social network about his investing acumen: We decided to help Intel in exchange for 10% of their shares. Is that too much or, too little? They were worth around 100 Billion Dollars when we made our offer. Now they are worth over 600 BILLION DOLLARS! Nine months, and they’ve increased in value over HALF A TRILLION DOLLARS. America’s stake is now over 60 Billion Dollars. When was the last time a President made America money?? Thank you for your attention to this matter! Recall that the second Trump administration decided to convert grants previously earmarked for Intel into a cash-for-equity swap. Regardless, if the President thinks that he has Midas’s brush, then the idea of securing (stealing?) a [portion of equity in major AI labs](https://www.cautiousoptimism.news/everyone-hates-ai-now/) could be more likely than we previously thought. Intel shares rallied on the ~news. The company is now up 237% this year alone. - Never interrupt your enemy in the middle of making a mistake, but does anyone have a good idea for how to keep the government *from*trying to abscond with 10% of OpenAI, Anthropic, SpaceX, and other major AI labs? - What about hyperscalers? Does the USG want a stake in Alphabet? Or just a JV spinout of the Gemini team? I don’t know, and I doubt that anyone knows, because moving a chunk of critical corporate power inside the state is a dumb-as-bricks idea. **Cybersecurity? **While the world tries to sort out who will get access to Anthropic’s Mythos-class models, famed for their cybersecurity chops, the hyperscalers are not sitting still. Amazon [announced AWS Continuum](https://aws.amazon.com/blogs/security/introducing-aws-continuum-security-at-machine-speed/) this week, a new model-agnostic harness for security agents. Described as “security at machine speed,” Amazon argues that the old model of handling enterprise cybersecurity through a loop of collecting, storing, and querying telemetry later displayed in dashboards is too slow. Correct. Yesterday, we [took a quick look at Ent Security](https://www.cautiousoptimism.news/mistral-gets-its-moment/), a startup building an agent that helps companies stop breaches and hacks by halting risky actions before they occur. There’s more to report, thankfully. Companies like Microsoft with its [MDASH ](https://www.microsoft.com/en-us/security/blog/2026/05/12/defense-at-ai-speed-microsofts-new-multi-model-agentic-security-system-tops-leading-industry-benchmark/?utm_source=chatgpt.com)(“*Microsoft Security multi-model agentic scanning harness*“) product, Palo Alto Networks and Databricks cooking *en couple*, Zscaler’s [improving AI-Guardian](https://www.zscaler.com/press/zscaler-expands-ai-guardian-uniting-tech-giants-secure-enterprise-ai-scale)— the list goes on. Mix in upcoming cybersecurity models from Mistral and [OpenAI’s own progress on the front](https://openai.com/index/gpt-5-5-with-trusted-access-for-cyber/), and maybe there’s enough being built to keep companies (and ourselves) secure in the AI era despite government fumbling. (Yes, this may be cope.) - Which cybersecurity-focused startups are growing the quickest? Noma Security, Cyera, Upwind, and Chainguard are all names to keep in mind. **Token terror:** Microsoft’s Copilot Cowork product — think Cowork for Office — dove into [early access in March](https://www.microsoft.com/en-us/microsoft-365/blog/2026/03/09/copilot-cowork-a-new-way-of-getting-work-done/). After adding Claude Opus 4.7 support in April, plugins and Ous 4.8 in April, Copilot Cowork is [now generally available](https://www.microsoft.com/en-us/microsoft-365/blog/2026/06/16/copilot-cowork-is-now-generally-available/), with Microsoft saying that it “is the fastest growing feature in the history of [its] Frontier program, [and] among the highest user satisfaction of any Copilot or agent experience [it has] shipped.” Cool. But it may not be cheap. Customers must have a Microsoft 365 Copilot User Subscription License *and *pay for intelligence consumed. This is why Microsoft made waves this week by [telling Axios](https://www.axios.com/2026/06/16/microsoft-copilot-cowork-tokenmaxxing-cowork) that it is “exploring a fine-tuned version of DeepSeek V4, or another open-source model, as a lower-cost alternative to the Anthropic and OpenAI models now powering its Cowork tool. *Hot damn*. Now, Microsoft is building its own foundation models, so in time it may have a high-performance, low-cost model to bring to bear on its own services (ala Cursor and its Composer model family). Today? No such model exists, and Microsoft wants its Office 365 customers to become addicted to using AI where they already work. So! Redmond needs to be able to offer gobs of intelligence at a price its customers will bear, and it can still generate margin *from*. Enter open-weight AI. - Note that DeepSeek is [still not on the American trade blacklist](https://www.reuters.com/world/china/us-holds-off-blacklisting-chinas-deepseek-more-than-100-firms-deemed-security-2026-06-17/), viz Chinese tech companies; that could change, but for now, its products are more usable in the United States than you might have thought. Why might Microsoft choose DeepSeek’s V4 model for Copilot Cowork? It’s insanely low cost. Currently the [14th-smartest model in the world](https://artificialanalysis.ai/) — at least until V4.1 drops — the DeepSeek model is the second-lowest-cost for the intelligence it offers. That makes it a screaming deal. Why *not *use it to help Copilot Cowork customers feel like token trillionaires? - The fact that Microsoft is considering a Chinese model at all is big news; that it is doing so because even the software giant can’t afford to allow its customers to go hog-wild in their AI usage really does say something. - Two things, actually. Microsoft has tooth/nailed itself to real enterprise AI adoption (hence the cost concerns), and we may finally be seeing the high (rising?) cost of truly frontier model usage become painful enough that *large*companies are looking elsewhere for*good enough and cheap*intelligence. - Three things, actually: [Reflection AI](https://reflection.ai/), got anything cooking that American companies can use? [📉](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) [Retail sales in China](https://www.cnbc.com/2026/06/16/china-economy-may-retail-sales-industrial-output-fixed-asset-investment-.html)…[oil prices next year](https://www.cnbc.com/2026/06/18/opec-iea-supply-glut-forecast-critical-strait-hormuz-reopens-oil.html)? …[Trump’s reviews](https://www.theamericanconservative.com/lessons-from-trumps-reckless-iran-war/)…[consumer electronics affordability](https://techcrunch.com/2026/06/17/ai-is-hurting-apple-in-more-ways-than-one-it-may-force-iphone-price-increases/)…[Trump-Bibi](https://newrepublic.com/post/212048/israel-cant-believe-trump-capitulation-iran-deal)…[working in media](https://www.ft.com/content/43801254-9be8-4db4-9742-c00a21d38ac0)…[long-term human intelligence](https://www.searchenginejournal.com/tiktok-shows-3x-more-ai-slop-than-youtube-report-finds/579521/)…[AI affordability](https://www.amd.com/en/products/processors/desktops/ryzen/ryzen-ai-halo.html)…[Waymo in construction zones](https://techcrunch.com/2026/06/18/waymo-recalls-nearly-4000-robotaxis-to-stop-them-driving-into-highway-construction-zones/)… **The free-wheeling days of AI:** Sometimes it’s worthwhile to compare tech companies to nation-states. *Microsoft As Country*, I suppose. The analogy fails as tech companies are bound by the laws of the nations where they operate, don’t issue currency, levy taxes, or field armies (though they do have access to aircraft, of course). We compare them, instead, thanks to the massive power and wealth that they control. Much like national governments. That’s why it wasn’t a big surprise that the recent G7 meeting included major AI leaders (Google DeepMind’s Demis Hassabis, OpenAI’s Sam Altman, and Anthropic’s Dario Amodei, among other tech execs). The companies (Alphabet, OpenAI, Anthropic, and, implicitly, Microsoft and Amazon through their partial ownership of the two AI labs) were on site to chat AI policy and make suggestions. Amodei called for democratic nations’ unity in the face of rapidly advancing AI technology, warning that the collected political leaders must “resist the temptation to splinter” in the face of new models like Anthropic’s own Mythos family. The three tech leaders “called for US-led collaboration on development of AI models,” [the FT reports](https://www.ft.com/content/573925dd-6d41-4185-810d-2b848195903d?syn-25a6b1a6=1). - Yes, it’s weird that Anthropic — currently both a company [too dangerous for the US government to use](https://www.cautiousoptimism.news/the-dod-is-trying-to-kill-the-next-great-american-company/), and[too dangerous for](https://www.cautiousoptimism.news/the-anthropic-fable-mess-explained/)to use as well — was there pitching policy in the same building as POTUS. But, hey, nation-state status grants you a little pull!*others* The idea of a democratic AI alliance is brilliant. The Internet is already shattering along authoritarian/democratic lines; why not cut to the chase and create an in-group for AI built in nations with real popular representation, and leave the autocrats out in the cold? - That would be, what, a sort of AI NATO? First NATO was so good, why not have a [second NATO](https://www.youtube.com/watch?v=zcGby-x5taA)? - NAITO? **Blocking and slowing data centers is a bipartisan affair** Look around the world as two forces collide: Market forces driving a global rush in data center construction, and political response to the same gold rush. You can find the point of collision in Africa, where a Microsoft data center in Kenya ran into power problems (amongst other issues). Power and water issues are cropping up in Mexico’s ‘data center valley.’ The UK and Denmark are grappling with how to handle rising power demands. Hell, a Spanish city [banned new data centers](https://www.datacenterdynamics.com/en/news/spains-city-of-lleida-bans-data-centers) after deciding that their long-term economic benefits outweighed their costs. ### Subscribe to Cautious Optimism to unlock the rest. Become a paying subscriber of Cautious Optimism to get access to this post and other subscriber-only content. [Subscribe](https://www.cautiousoptimism.news/subscribe/?utm_source=Cautious+Optimism&utm_medium=website&utm_campaign=article-paywall)