Bloomberg Tech
2026-06-16 · Hosted by Caroline Hyde, Ed Ludlow · Bloomberg / iHeartMedia
Executive Summary
The U.S. Commerce Department ordered Anthropic to block foreign-national access to its two most advanced models, Fable 5 and Mephos 5, over a “jailbreak in the code” security concern, prompting Anthropic to pull both products worldwide while it negotiates — an unprecedented step that competitors like Cohere say has triggered a “huge number of inbounds” from governments and enterprises seeking to diversify. Markets were broadly upbeat: the Nasdaq 100 rose nearly 3% toward record highs, buoyed by a feel-good SpaceX IPO whose shares gained up to 10% on day two (market cap ~$2.32 trillion) and optimism over a U.S.–Iran interim deal to reopen the Strait of Hormuz. In capital-markets news, Nvidia is seeking to raise at least $20 billion in its first bond sale since 2021, while Fox agreed to buy Roku at a $22 billion enterprise value and Salesforce bought AI firm Thin for $3.6 billion.
Key Stories & Changes
1. Government Forces Anthropic to Disable Advanced AI Models
The Commerce Department issued a directive Friday at 5:21 PM ET concerned about foreign nationals accessing Fable 5 and Mephos 5 — the key products ahead of Anthropic’s near-trillion-dollar IPO.
Anthropic chose to deny access worldwide to both models rather than operate under restrictions it deemed “unworkable,” including for foreign nationals residing in the U.S.
The move marks a break from Trump’s executive order of a week earlier, which stated it should “not be interpreted as establishing a licensing regime.”
The cited issue is a “jailbreak in the code” — circumventing guardrails meant to prevent the model from being used for cyber-security attacks. Anthropic disputes there is a “universal jailbreak” and believes the government is overstating or misunderstanding the risk.
Irony noted: Two days before, Anthropic’s leader said the government should have the ability to ban risky models.
An Anthropic executive met at the Commerce Department today; CEOs of OpenAI, Anthropic, and Google are slated to attend the G7 summit in Geneva, with a face-to-face Trump–CEO meeting Wednesday.
2. Industry Reaction — Cohere and the Open-Source Argument
Cohere (Canadian frontier AI firm) reported a “huge number of inbounds” from business customers and governments outside the U.S. and China seeking to diversify their tech stack.
Chief AI Officer Joelle Pineau noted only four countries can train foundation models today: U.S., China, France, Canada.
The episode underscores global concern over over-reliance on U.S. technology and tech sovereignty, especially as the EU examines the issue.
3. Nvidia Returns to the Bond Market
Nvidia is seeking at least $20 billion in its first bond sale since 2021 — seven tranches, maturities 2 to 30 years, for general corporate purposes including refinancing.
Despite ~$200 billion in expected free cash flow over the next year, BI’s Robert Schiffman says the rationale is funding strategic partners (e.g., OpenAI, possibly SpaceX) and exploiting a near-zero weighted-average cost of capital on a $5 trillion equity cap.
4. SpaceX Day Two & the “Race to Capital”
SpaceX shares rose up to 10% on day two, reaching a ~$2.32 trillion market cap; the company sold 83.3M extra shares via an instant green-shoe, raising more than $75 billion (cited later as $85.7 billion).
JPMorgan’s Stephanie Aliaga said ~$1.5 trillion in new shares are coming to market this year (per JPMorgan analysis), reversing two decades of ~$12 trillion in buyback-driven share shrinkage.
A six-month mark is flagged as the likely point of greater rebalancing volatility.
5. Media & Software M&A
ROKU: Roku — Acquired at $22B EV — Fox buys Roku to become a neutral aggregation platform; third-largest US TV player by viewership
FOXA: Fox — Down (cash & stock deal) — Investors wary of stock-financed deal and strategy shift
CRM: Salesforce — Up ~1pp — Buys AI firm Thin for $3.6B to bolster agentic capabilities
6. China Tech: “Source of Funds”
Citi’s Alicia Yap argues global investors treat China internet as a “source of funds,” selling local names to fund the hardware AI trade in Korea, Taiwan, and Japan.
China models (Alibaba’s Qwen 3.5→3.7, Tencent’s Hunyuan 3, MiniMax) are “increasingly closing the gap” with frontier models; Chinese AI maker GLM/Jipu surged after a JPMorgan price-target hike.
7. Apple’s Revamped Siri & Data Center Frontiers
Bloomberg’s Mark Gurman called the new Siri “just good enough,” on par with recent ChatGPT/Gemini/Claude for ~95% of users.
Former Meta CTO Mike Schroepfer (Gigascale Capital) argued ocean/submarine data centers are ~100x cheaper per ton of mass than orbital ones, which only SpaceX can economically pursue (earliest orbital deployment 2028).
Trends Identified
1. Government Intervention as a New AI Risk Vector
The Anthropic directive establishes that even the most capable models can be pulled from the market overnight on security grounds. This injects a new category of volatility — political and regulatory access risk — into the AI value chain, and is driving enterprises and foreign governments toward diversification and on-premise control.
2. The “Race to Capital”
What began as a race to IPO (SpaceX, then Anthropic and OpenAI) has pivoted to a race to raise capital across every instrument: SpaceX’s record equity raise, Nvidia’s bond return, and large equity offerings from Alphabet and Meta. Roughly $1.5 trillion in new shares this year reverses a two-decade trend of share scarcity, testing the market’s absorption capacity.
3. Tech Sovereignty and Open-Source Momentum
The forced shutdown amplifies concerns about over-reliance on a small group of U.S. labs. With only four countries able to train foundation models and Chinese open-source models closing the gap, the episode is accelerating interest in diversified, controllable, and open alternatives.
4. AI’s Insatiable Capital and Infrastructure Demand
From Nvidia funding partners to debates over submarine versus orbital data centers, the dominant theme is that the AI build-out is becoming dramatically more expensive and multi-year, forcing selectivity as company financials and leverage begin to diverge. —-
Sentiment Analysis
Overall Market Sentiment: Cautiously Bullish
Equities pushed toward records on SpaceX enthusiasm and Iran-deal optimism, but the Anthropic episode and capital-supply concerns kept an undertone of caution.
Risk Factors Highlighted
Regulatory access risk: Governments can force advanced AI models off the market overnight, as with Fable 5 and Mephos 5.
Capital absorption: ~$1.5 trillion of new equity plus large bond raises could strain the market, with volatility likely around the SpaceX six-month mark.
Tech sovereignty / over-reliance: Global dependence on a few U.S. labs exposes other economies to abrupt access loss.
China gap-closing: Chinese models and open-source alternatives erode the U.S. lead and Washington’s control.
SpaceX rebalancing mechanics: Index inclusion and passive-fund rebalancing could drive volatility as float increases.
M&A execution / dilution: Fox’s stock-financed Roku deal pressured shares amid neutrality and strategy concerns.
Iran deal fragility: Interim, phased agreement with unresolved Strait of Hormuz fee and economic-relief questions.
Leverage divergence: AI investment is turning many hyperscalers free-cash-flow negative, diverging from cash-rich Nvidia.
This episode was covered in today’s The Market Signal — 2026-06-16, a cross-source synthesis of multiple podcast reports.