Apple Hikes the 16-inch MacBook Pro $300 as Asia Clones Washington's Gated Mythos

01Tim Cook called Apple's pricing "unsustainable." The 16-inch MacBook Pro just went up $300

Tim Cook described Apple's pricing as "unsustainable," then said the increases were "unavoidable." The 16-inch MacBook Pro went up $300. The 11-inch iPad Air climbed from $599 to $749, a jump of $150. Even the HomePod Mini took a $30 bump, to $129.

Those numbers land on buyers who already pay a premium for Apple hardware. Cook offered the language of inevitability, not apology. The Verge put the question more bluntly in its report: why should customers pay more for Big Tech's AI spending?

The framing follows the money. Apple, like every large platform, is directing capital toward AI infrastructure while its consumer devices carry higher sticker prices. The company has not broken out how much of any single increase reflects that buildout. What it has done is raise prices across the lineup and tell customers the prior model could not hold.

Cook is squeezing one end of the business while the other end leaks talent. Paul Meade, the Apple vice president in charge of the Vision Pro headset, is reportedly leaving to join OpenAI's hardware team. He ran Apple's most expensive recent hardware bet, a $3,499 headset that never reached mass volume.

Meade's destination is the sharper detail. OpenAI is assembling a hardware group, and it is hiring the executive who shipped Apple's spatial computer. Apple now faces a rival that is buying both the engineering leadership and, indirectly, the consumer wallet share that funds the same race.

The two pressures connect at the customer. Apple raises prices to absorb the cost of competing in AI, then watches a senior hardware leader walk to the company defining what that competition looks like. One side of the ledger asks buyers to pay more. The other side hands a competitor the person who knows how Apple builds new device categories.

Apple has not said whether further increases are coming, and it has not commented on Meade's reported departure. The HomePod Mini sits at $129, the iPad Air at $749, the MacBook Pro $300 higher than before. The company's message to customers is that these prices reflect what the business now costs to run.

02Washington Gated Mythos to 100 Vetted Institutions. Asia Is Selling an Unlicensed Copy.

A new class of AI models is shipping across Asia with one feature American labs can no longer offer: no export license required. The startups behind them promise Mythos-like capabilities, according to TechCrunch, and they sell to anyone. U.S. labs may never recover the market, the report said.

That pitch lands two weeks into a standoff over who can touch Anthropic's flagship model. On Friday, Commerce Secretary Howard Lutnick wrote to Anthropic chief compute officer Tom Brown, lifting the government's block on Claude Mythos 5. Access goes to "certain trusted partners." More than 100 U.S. companies and agencies made the list, according to Semafor.

The list is the point. Deployers outside those 100-plus institutions cannot run Mythos legally. Even inside them, non-American employees need authorization to use the model, according to TechCrunch. Lutnick cited "appropriate safeguards" and said Anthropic committed to work with the government on "protocols and standards and releases."

The block followed warnings from Amazon and other companies that Mythos and its weaker cousin Fable 5 could be "jailbroken" for malicious purposes, according to Semafor. Both models shut down. Fable, briefly the most powerful model widely available to consumers, stays blocked; people close to the talks said a release is moving forward, though they gave no timeline.

While Washington decides which partners qualify, the Asian models carry no such gate. A developer who cannot get on the trusted list, or who employs engineers the U.S. will not clear, now has somewhere else to go. Whether those models actually match Mythos is unverified. The startups only promise that they do.

Anthropic gained roughly 100 sanctioned customers on Friday. What it cannot sell to sits outside U.S. borders, where the competition ships without a letter from the Commerce Department.

03The fix for AI's power bill now ranges from Earth orbit to a claimed 1,000x cut

The scarce input for the next wave of AI is no longer silicon. It is electricity, and two ventures surfaced this week proposing opposite escapes from the same ceiling.

The first plan moves the data center off the planet. Elon Musk has promoted putting compute in orbit, where sunlight is constant and the vacuum carries heat away. SoftBank's CEO has questioned that vision, according to TechCrunch, and the report makes clear he is not the only skeptic. Others in the piece press the same unanswered points: launch mass, radiation hardening, and how anyone services a failed rack 500 kilometers up. No costed orbital plan has been published.

The second plan keeps the racks on the ground and attacks consumption instead. Databricks' former AI chief has started a company that he says can cut AI's power bill by a factor of 1,000, per TechCrunch. That number is the founder's claim, not an audited result. The reporting ties it to an image-generation tool the startup uses to show its technology can replicate what conventional AI systems already do, at a sliver of the energy.

Both bets concede the same premise. Frontier training and inference now collide with the power supply before they collide with chip availability, so the open question has become how to feed the load rather than how to compute it. One answer chases new generation in places the grid does not reach. The other tries to make the existing draw smaller by orders of magnitude.

Neither has shipped proof. Orbital compute remains a pitch with named doubters and no public unit economics. The 1,000x figure remains a founder's projection attached to a demo, not a deployment. What both reduce to is a wager on which lever moves first: supply, by hauling solar capacity above the atmosphere, or demand, by rebuilding how models burn watts. Whichever clears, the binding number for the next phase of AI scaling sits on the utility bill, not the benchmark sheet.