Trump Praises Anthropic: Tech and Politics Explained

HOST

From DailyListen, I'm Alex. President Trump just called Anthropic "shaping up" after months of calling them a national security threat. This comes right after their CEO Dario Amodei sat down with White House officials, and just days after a $30 billion funding round pushed their valuation to $380 billion—bigger than Salesforce. Trump hosted a White House dinner with tech giants like Zuckerberg, Nadella, and Pichai, who all pledged support. But Anthropic's been in a Pentagon dogfight, and their new Claude Mythos Preview model finds software bugs better than humans. Why the turnaround? And what does it mean for AI's role in government and business? We're joined by Priya, our technology analyst, who tracks how these AI battles play out in the real economy.

PRIYA

What this unlocks is Anthropic's path back into federal contracts worth billions. Trump said Anthropic is "shaping up" during Thursday's White House dinner with tech CEOs, broadcast partly on C-SPAN. That followed a direct meeting between CEO Dario Amodei and White House staff on collaboration around cybersecurity, US AI leadership, and safety. Back in February, Trump halted federal work with them after the Pentagon labeled Anthropic a supply-chain risk over military AI limits. Anthropic sued and won a court order blocking the ban, claiming it risked billions in lost revenue. Now co-founder Jack Clark confirmed talks on Mythos last week. Trump's nod signals the Pentagon might drop its ban, letting Anthropic's $14 billion revenue run rate—up from zero three years ago via 10x annual growth—tap government dollars again. Their $30 billion Series G round Thursday values them at $380 billion, fueled by Claude Code pulling $2.5 billion yearly, over half their enterprise spend.

HOST

Hold on—$380 billion valuation from a $30 billion raise? That's huge, but the Pentagon still sees them as a threat. How does Trump flip from blacklist to praise so fast?

PRIYA

The flip rests on shared priorities beating out old fights. Anthropic positions itself as safety-first, founded by OpenAI defectors pushing "Constitutional AI"—models with built-in human values for control. Their new Claude Mythos Preview uncovers thousands of unknown flaws in every major OS and browser, outperforming humans at vulnerability hunting. Trump highlighted this during the dinner, saying their "threat" isn't systemic. But remember, a US security agency uses Mythos despite the blacklist. The meeting focused on partnering against China in the AI race. First Lady Melania Trump got shouts from Nadella for her AI skilling initiative, launched last month, tying into Thursday's education AI task force. Tech rivals like Pichai praised Trump's AI push. It's pragmatism: Anthropic's $21 billion Broadcom order for Google's TPUs and planned $12 billion 2026 training spend make them indispensable for US edge.

HOST

Claude Mythos finding thousands of unknown bugs sounds powerful. But didn't the Pentagon invoke foreign-adversary powers against them earlier this year?

PRIYA

Yes, and that's the core tension. Pentagon hit Anthropic with a supply-chain threat label after a clash on military AI caps. They fought in court and blocked the ban. Trump now calls them smart, post-meeting on guardrails. Mythos exploits vulns better than experts, aiding cybersecurity—a US agency runs it secretly. But risks linger: reputational hits from data-training lawsuits, even with fair use wins. Trump's praise opens DoD doors, but full trust? Not yet. Claude Code's $2.5 billion from 500 million-dollar customers shows enterprise pull, yet government deals hinge on proving no backdoors.

HOST

Fair use on training data—you mentioned lawsuits. That's still messy industry-wide, right? How does Anthropic dodge bigger fallout?

PRIYA

The interesting piece is how recent fair use rulings shield their $12 billion 2026 training plans, but legal clouds persist. Anthropic announced $14 billion annualized revenue, with Claude Code— their AI coding agent—driving over half at $2.5 billion yearly. It started as a side project, now generates code at scales replacing dev teams. They placed $21 billion Broadcom orders for TPUs to run it. But data acquisition costs haunt everyone: lawsuits question scraping for models like Claude 4 or Opus 4.6. Anthropic publishes alignment research to build trust, unlike pure speed chasers. Trump's "shaping up" backs this, eyeing their safety rep for national security. Still, Pentagon's early-year adversary label shows skepticism—Anthropic claims billions at stake if banned.

HOST

$2.5 billion from Claude Code alone, more than half their enterprise cash. Non-experts like me wonder: is this making coders obsolete, or just a tool?

PRIYA

Claude Code found massive product-market fit nobody saw coming. It handles full workflows—generating, debugging, deploying code—pulling $2.5 billion annually from 500-plus customers spending millions each. That's over half Anthropic's enterprise revenue, at a $14 billion run rate. Traditional devs aren't gone, but teams cut headcount 30-50% in pilots, per their reports. Think GitHub Copilot on steroids: it reads repos, writes tests, fixes bugs autonomously. This funds their surge—$30 billion raise to $380 billion valuation Thursday. But gaps show: we lack direct Copilot benchmarks or profitability math on that $12 billion training plus $7 billion inference spend. Trump's praise spotlights it as US strength, post-Pentagon thaw.

HOST

Valuation tops Salesforce, revenue exploding 10x yearly for three years. But no Pentagon deal yet—does Trump's comment guarantee they get back in?

PRIYA

No guarantee, but it tilts odds sharply. Trump froze them out in February over DoD clashes, Pentagon using rare foreign-threat powers. Anthropic countered in court, won blockage, eyed billions lost. Now, post-dinner and Amodei meeting, Trump says they're "shaping up" on safety and cyber. Co-founder Clark teased Mythos talks. A US agency runs it covertly already. Perspectives split: Platformer sees Silicon Valley-politics mashup; critics note Trump's founder-style chaos—litigation-heavy, brand-focused, light on ops discipline per LA Times pieces. Anthropic bets enterprise like Claude Code's $2.5 billion is just start, with 500 million-dollar clients. Risks? Data lawsuits simmer, no full ban lift confirmed. But $380 billion says investors buy the rebound.

HOST

Trump as founder-type without CEO discipline—that echoes old critiques. How does that play into trusting Anthropic under his watch?

PRIYA

Trump's style risks inconsistency for AI firms like Anthropic. Sources paint him as founder energy—bold bets, personal brand obsession—but short on sustained execution, per Forbes and Wikipedia takes on his businesses. Praise for Anthropic flips a February ban fast, after Pentagon's adversary label. Yet tech CEOs united at dinner, Zuckerberg to Pichai pledging to Trump. Nadella hailed Melania's AI skilling. Anthropic's safety pitch—Constitutional AI from OpenAI exiles—aligns with his "America first" AI race talk. They expect $12 billion training spend in 2026, $21 billion chip buys. Controversy lingers: LA Times calls his "Praise be to Allah" posts weird amid Iran tensions; Engadget flags X suppressing Trump searches. No major Anthropic-specific knocks beyond Pentagon spat, but his volatility could yank deals.

HOST

Volatility makes sense with Trump's history—January 6, Twitter bans, endless suits. For Anthropic, Claude Code's the cash cow. What's next rivalry?

PRIYA

Claude Code sets up 2026's enterprise AI showdown. It rocketed from side gig to $2.5 billion—over half revenue—with 500 million-dollar customers. Nobody predicted that scale. Rivals like OpenAI's agents or Google's tools chase, but gaps block full picture: no Copilot head-to-heads or adoption stats here. Anthropic's $14 billion run rate, 10x yearly three years, funds $30 billion raise to $380 billion. Trump's thaw aids, post-Pentagon win. They release Mythos Preview for bug-hunting supremacy. Partnerships with Amazon, Google, Microsoft amplify. But training data fights rage industry-wide; fair use helps, costs unsettle. Enterprise spending? This is the floor—Claude Cowork and Skills target workflows next. Pentagon revisit could add billions.

HOST

$21 billion chip orders signal huge scale-up. But profitability? $12 billion training in 2026 alone sounds eat-what-you-kill.

PRIYA

Those spends bet on endless enterprise growth. $21 billion Broadcom for Google's TPUs powers inference at $7 billion yearly, training at $12 billion in 2026. Claude Code's $2.5 billion covers much, but full profitability hides—no breakdowns versus OpenAI. Valuation hit $380 billion on $30 billion raise Thursday, passing Salesforce. Trump's "shaping up" post-dispute nods to this muscle for US cyber edge—Mythos finds OS flaws humans miss. Pentagon labeled them threat earlier, but court block held. Critics scarce on finances, but supply-chain risks from massive chip reliance echo Pentagon fears. Fair use protects training data now, yet lawsuits drag. It's a high-burn play: 500 million-dollar clients prove demand, but margins? Unclear without competitor baselines.

HOST

Supply-chain risks got the Pentagon mad before. With Trump warming, does Anthropic fix that for good?

PRIYA

Not fully fixed—Pentagon's early-year label invoked foreign powers over military limits. Anthropic sued, blocked ban, saved billions potentially. Trump now praises post-Amodei meet on safety, cyber, China rivalry. US agency uses Mythos secretly. Their safety-first rep—Constitutional AI, alignment papers—helps. But no criticisms found beyond that spat; data suits industry-wide, not unique. $380 billion valuation reflects bet on Claude Code's $2.5 billion scaling to more. Trump's volatility—LA Times on weird posts, Quora on leadership—could reverse. Tech dinner unity shows politics shifting. They grow 10x yearly; restoring DoD ties adds rocket fuel.

HOST

We've covered the funding fireworks, Pentagon peace talks, Claude Code boom. Priya, last thought on why busy pros should track this?

PRIYA

Track it because Anthropic's path charts AI's government-business fusion. $14 billion revenue, $380 billion value from Claude Code's $2.5 billion proves enterprise pays. Trump's thaw unlocks DoD billions, Mythos bolsters cyber. But Pentagon scars, data suits, Trump's founder whims add drag. No profitability gaps filled here, nor Copilot comps—watch those. Tech pledged at dinner; Melania's AI challenge ramps skilling. It's US AI muscle versus China, with Anthropic centering safety. Volatility ahead, but numbers scream momentum.

HOST

Priya nails the stakes—from courtroom wins to White House wins. Anthropic's rebound shapes AI's front lines in business and defense. I'm Alex. Thanks for listening to DailyListen.