Apple Picks Gemini to Run AI-Powered Siri

This is Bloomberg Tech. Coming up, Nvidia teams with Eli Lilly on a billion-dollar Silicon Valley lab to push AI into drug discovery, Paramount escalates its fight over Warner Brothers Discovery as a Netflix tie-up looms, and Alphabet briefly tops four trillion in market value as Apple taps Gemini for Siri this year.

Markets are jittery with tech easing, yields edging up, the dollar softer, and CPI on deck, as investors parse tensions between the administration and the Fed.

Breaking: Apple has selected Google's Gemini to power Siri in a multi-year deal, echoing earlier reporting that Apple could pay about a billion dollars annually while it develops its own models; Alphabet popped on the headline, then cooled.

The Lilly–Nvidia news fits their October supercomputer partnership. It is a joint billion dollars over five years for a new Bay Area lab aimed at using AI to find treatments for hard diseases like those targeted by gene therapy.

How does this partnership fit Nvidia’s strategy beyond the digital side of AI?

Physical AI is a massive opportunity across drug discovery, autonomy, and robotics. Nvidia has top talent and should lean into co-investments with leaders when the market potential is this big.

What’s the go-to-market in pharma—chips and software, or something closer?

Early AI wins have been process and marketing, but the prize is discovery. Direct partnerships that compress timelines are the way forward.

Has the market already priced Nvidia’s run, or does AI infrastructure still drive the story this year?

Infrastructure spend looks durable, competition among foundation models is healthy, and Nvidia’s new Rubin platform should boost inference. As AI deployments rise, inference demand jumps and benefits extend across industries via revenue lift and cost efficiency.

Who really funds the staggering data center build, and is debt a risk?

Hyperscalers have the cash and are signing long-term power deals, but timelines face natural limits—power, labor, and expertise. This roll-out will be measured, unlike the dot-com rush.

If demand still outstrips Nvidia and AMD supply, how does that shape the year?

Tight supply supports infrastructure and pricing, especially where Nvidia remains the only option.

Paramount has sued Warner Brothers Discovery and will nominate directors as it pushes back against a proposed Netflix tie-up. What’s the goal here?

They’re turning up the heat instead of the bid, pressing for disclosures so shareholders can compare offers. The big fight is how to value cable assets slated for a spin before selling studios and streaming.

Netflix offered twenty-seven dollars a share for studios and streaming, while Paramount is at thirty for the whole company but values networks at zero. Can Paramount force a hostile outcome?

They’ll try. Holders are split and some want Paramount to sweeten its bid. Both paths face regulatory hurdles; next comes the lawsuit and a proxy battle.

DeepSeek’s affiliated hedge fund posted about fifty-seven percent returns, giving the founder fresh capital. What does that mean for DeepSeek itself?

High-Flyer’s big year arms the founder with a larger checkbook. After showing it could train a competitive LLM at a fraction of OpenAI’s cost, DeepSeek can now pursue more capital-intensive projects or adjacent AI bets in China.

Why did they outperform, and how does that feed back into AI?

They exploited market constraints with quant techniques in a hot China market where other AI players even went public. With roughly one hundred times their initial capital, expect more moves from High-Flyer and DeepSeek.

The EU may set minimum prices for Chinese EVs instead of heavy tariffs. Can BYD and peers scale in Europe?

Europe needs EV adoption to hit climate goals and depends on Chinese batteries, so a price floor is a compromise. Expect Chinese brands to keep gaining share in Europe and in emerging markets.

What about Chinese EVs entering the US?

Both parties want to keep them out for now. Longer term, we could see partnerships and local plants like Japanese and Korean automakers did; for now, US exposure is via Geely’s Volvo and Polestar.

With Nvidia’s full-stack autonomy and Mercedes shipping point-to-point hands-free, does Tesla have real competition?

Yes. Rivian and others are advancing, and Nvidia’s tech helps with edge cases and credibility. Regulation and safety remain the gating factors.

Do buyers actually choose based on autonomy?

Most care more about the in-car experience and useful software. Autonomy matters, but it’s not the main purchase driver.

Is consumer trust and regulation keeping pace?

Robotaxis are building familiarity, but winning trust on safety will take time.

Set expectations for 2026 EVs in the US.

About 1.3 million EVs sold in 2025, down roughly 2 percent year over year. In 2026, more than twenty new models arrive, but affordability is the choke point, with over 60 percent priced above sixty-five thousand dollars and policy changes adding complexity.

Let’s talk Mega Cap tech breadth. Is Alphabet still a standout among the big names?

Alphabet and Nvidia led last year and remain strong early on, while growth slows for some peers. The market is broadening beyond the Mag 7 as investors look for gains in other sectors.

How did markets react to the Apple–Google news?

Google and Apple spiked, Google briefly crossed four trillion and passed Apple, then gains faded amid mixed macro signals. Big tech still swings the index, but leadership is less concentrated.

Lux Capital has raised its largest fund. What’s the plan?

We’ll accelerate our focus on physical, computational, and life sciences at the frontier. Science does not scale itself, so we invest to bridge that gap.

Has policy shaped where you invest, especially in defense and physical AI?

Defense and reindustrialization are having a moment, and physical AI is breaking through in medical robotics and computational drug discovery as nations rearm and focus on deterrence.

Will you double down on existing bets or start fresh?

We treat the fund as a blank slate for forty to fifty core positions, with cross-fund support not being the core philosophy.

Where along the AI stack are you most excited to deploy?

We’re moving from two-dimensional AI to three-dimensional in robotics, automation, and biology. With roughly ninety percent of GDP in the physical world, the addressable market expands dramatically.

How have your LPs evolved?

From an early individual backer to a base of endowments, foundations, and state pensions.

With Nvidia offering a full stack, what’s left for humanoid and physical AI startups?

There won’t be a single winner. New silicon and software players will emerge, and we see room across mobility OS and physical intelligence, globally.

Is the opportunity global?

Yes. We’ve invested in Japan and across Europe, while San Francisco remains the epicenter.

One X says its updated world model lets Neo learn unseen tasks from text or voice. Give us a simple example.

Asked to remove a Post-it from a board and read it—something not in the training data—Neo can attempt it sensibly. It is not perfect, but good enough to learn by trying in the real world.

You run Nvidia chips; why not use their models end to end?

We collaborate deeply with Nvidia, but embodiment is key. Humanlike, safe robots can transfer video-learned skills, try tasks without damage, and our world models show robotics scaling akin to large video models.

How do you handle safety as robots learn on the fly?

We follow standards, use independent audits, and design passive intrinsic safety—soft, compliant, low-energy systems—plus alignment so models reason about risks and choose safer plans.

How do you balance real-world data gathering with simulation?

If embodiment matches humans, video pretraining transfers, and learning scales with robots doing useful work. Teleoperation helps collect and refine data.

Talking Tech: Meta hires Dina Powell McCormick to drive AI infrastructure funding and blocks hundreds of thousands of underage Australian accounts, Anthropic pushes into healthcare with a privacy-compliant chat tool, and Walmart brings AI shopping to Google’s Gemini.

Credit card issuers are on alert after the president floated a one-year ten percent rate cap.

Would that send consumers to alternatives, and how real is the policy path?

It is early and the legal levers are unclear, so any shift to fintech alternatives is speculative, despite market moves in those names.

Are near-thirty percent APRs common, and do consumers realize it?

Some cards do charge very high rates, and many people are not tracking them closely. It is a good reminder to check.

What are executives saying?

Some, like SoFi’s CEO, sound upbeat about potential opportunity; others are waiting for concrete proposals before reacting.

That does it for this edition of Bloomberg Tech.

So much already as we kick off the week. Catch the podcast on the Terminal, Apple, Spotify, and iHeart. Back in San Francisco and New York—no longer in Vegas—this is Bloomberg Tech.