BUSINESS

Oil prices on Tuesday extended their increase from morning action after the U.S. military said it struck Iranian targets in the Strait of Hormuz in response to Tehran’s overnight attacks on ships transiting the vital waterway.

Pink Floyd’s blockbuster ‘Wish You Were Here’ simultaneously returns to three charts in the U.K.: the Official Albums Sales, Official Physical Albums and Official Album Downloads rankings.

The best Rotten Tomatoes scored movie of 2026 when you combine critic and audience reviews is a must-watch, and was overlooked at the box office.

TheStreet aims to feature only the best products and services. If you buy something via one of our links, we may earn a commission.

Why we love this deal

It’s no secret that Labor Day offers some of the best discounts of the year on big-ticket items, with deals comparable to only those of Black Friday. Generally, we think about snagging savings on furniture, appliances, and premium fashion during this holiday sales event, but with Walmart’s Beauty Event, happening now until Friday, September 4, it’s also the perfect time to stock up on high-end hair tools from Dyson, dermatologist-approved skincare, and luxury fragrances for less.

One of our favorite deals is on the bestselling and highly rated Dolce & Gabbana Light Blue Summer Vibes Eau de Toilette, which is currently 49% off at Walmart. Instead of paying the regular price of $70, you can score the generous 3.3-ounce bottle for just $36. This limited-edition release is no longer in production, so once it sells out, this beloved floral and woody fragrance with notes of bergamot, peach, and cedar will be gone forever. 

Dolce & Gabbana Light Blue Summer Vibes Eau de Toilette, $36 (was $70) at Walmart

Courtesy of Walmart

Shop at Walmart

Why do shoppers love it?

Over 800 shoppers have given this designer perfume a perfect five-star rating. One shopper called it, “The most beautiful-smelling perfume ever.” The eau de toilette was designed by French perfumer Olivier Cresp, the nose behind many iconic scents. The lightweight fragrance is said to capture the feeling of a romantic escape to an Italian coast by blending the fresh notes of Calabrian bergamot with sweet, ripe peaches and a soothing woody base. The earlier reviewer continued to rave, “It smells so expensive, and it lasts long. I get so many compliments — it’s such a beautiful fragrance.”

Related: Walmart has a 18-karat white gold tennis bracelet for 88% off

The light and airy scent is perfect for wearing to the office, dinner with friends, or whenever you want to feel fancy, even if that’s just running to the post office. As one shopper explained, “It’s not overpowering, just a beautiful scent.” You’ll want this scent in your collection, and not just because of its fantastic fragrance. It comes in a beautiful glass bottle that you’ll want to display on your shelf. The classic majolica print gives a nod to the Italian seaside with a rich Capri blue reminiscent of the Blue Grotto waters. The same reviewer praised the design: “The bottle is a work of art.”

Details to know 

Size: 3.3 ounces.

Fragrance type: Eau de toilette.

Notes: Bergamot, peach, and cedar.

Average shopper rating: 4.5 out of five stars.

This specific formula is designed for women, but the Light Blue Summer Vibes was also released in a men’s version. The Dolce and Gabbana Light Blue Pour Homme Summer Vibes comes in a slightly larger 4.2-ounce bottle, available for $42 at Walmart.

Shop more deals

Dolce & Gabbana The One Eau De Parfum Spray, $55 (was $75) at Walmart

Versace Bright Crystal Eau De Toilette, $47 (was $76) at Walmart

London by Burberry Eau De Parfum, $34 (was $53) at Walmart

Take advantage of Walmart’s Beauty Event deals by adding the Dolce & Gabbana Light Blue Summer Vibes Eau de Toilette to your shopping cart while it’s on sale for just $36. Once this fragrance sells out, it will be gone for good, so don’t miss your chance to make it part of your collection. 

The American economy is growing, but the footing looks a lot less comfortable.  

GDP expanded at a 1.5% annualized pace in Q2, down from 2.1% in Q1, while July payrolls dropped by 23,000 and unemployment held near 4.1%, as reported by CNBC.

Yet private domestic demand jumped to 4.2%, as reported by Investing.com, suggesting that the engine hasn’t stalled. Goldman Sachs CEO David Solomon argues a bigger story is developing beneath these mixed signals.

Consumers are at the heart of that dichotomy.

Personal income increased by 0.4% in July, but real spending was mostly flat, the saving rate tanked to 3%, and retail sales fell 0.6%. Inflation also remains sticky, with the PCE index running at 3.7% above last year, as reported by Reuters.

Then we have the enormous AI investment cycle. Companies continue to borrow heavily for funding infrastructure, raising questions about whether the boom might eventually create a credit problem or lead to a painful reset.

Solomon’s remains unexpectedly constructive. 

In a CNBC interview, he talked about how he sees enough strength beneath the surface to look past the current risks.

His more striking argument is that a familiar technology might unleash an “extraordinary” productivity boom that could potentially revamp America’s long-term growth potential. 

Why Solomon sees a stronger U.S. economy beneath the noise

Solomon argues that even though the current headwinds might slow down the economy, he doesn’t believe they have broken its engine.

That distinction effectively shapes how he interprets every major risk ahead. 

More Economy:

Kevin O’Leary raises stark concern about inflation

Goldman Sachs delivers its verdict on inflation and jobs

Bank of America issues stark warning on Fed and economy

“Generally speaking, you know, the consumer is still pretty resilient,” Solomon said. “The economy is performing well.” 

It’s important to note that consumer spending remains the biggest source of U.S. activity. If households continue spending while labor conditions cool off, the economy could absorb weaker patches without tipping into a contraction.

The second pillar is the investment cycle.

Solomon pointed to an “enormous investment cycle” contributing to growth and activity, a nod to the heavy spending surrounding AI, data centers, and related infrastructure. According to him, the capital buildout is supporting demand today, even before the promised productivity gains arrive.

Corporate earnings reinforce the argument.

Solomon called the numbers coming in as “extraordinary,” describing it as “a big tailwind for the market and for the economy.” Rising profits offer businesses greater capacity to invest, hire, service debt, and absorb higher borrowing costs.

The broader data support that argument. 

U.S. corporate profits from current production shot up by $400.9 billion in Q2, which is more than five times the $74.4 billion increase recorded in Q1.

Moreover, FactSet’s latest analysis showed that the Magnificent 7 stocks delivered 118.5% earnings growth in Q2, while the other 493 S&P 500 companies posted blended growth of 31.8%, their strongest pace since late 2021. 

Nevertheless, Solomon didn’t predict a frictionless outlook, either. 

He acknowledged “bumps and headwinds” from the Middle East, arguing that the trade policy and tariffs remain “a headwind to some degree.” Nevertheless, he sees these as obstacles that the economy can efficiently work through.

Perhaps the long-term case is a lot more ambitious. 

As AI moves into enterprises, Solomon forecasts an “extraordinary” productivity boom that could potentially create “a fundamentally higher growth rate.” 

Goldman Sachs CEO David Solomon believes the current headwinds impacting the economy haven’t broken its engine.Brendon Thorne/Bloomberg via Getty Images

AI productivity boom is showing up, but only in pieces

Solomon’s AI optimism isn’t entirely built on promise. 

We’re seeing some early evidence that the technology is saving time, elevating less-experienced workers and spreading quickly. The bigger question, though, is whether those isolated gains could become an economy-wide acceleration.

The bull case starts with the numbers.

U.S. nonfarm business productivity rose 2.2% year over year in Q2 2026. Since late 2019, it has risen at a 2.1% annualized rate, above the 1.5% pace of the previous business cycle. Though it’s unfair to attribute the lion’s share of that improvement to AI, the timing is consistent with an emerging contribution.

Moreover, workplace evidence is a lot more direct.

St. Louis Fed data showed that 39.2% of employed adults used generative AI at work by Q2, up substantially from 28.2% in Q3 2024. AI-assisted hours increased to 6.3% of total work time, while reported time saved reached 2.2%.

Separately, an NBER field study revealed that AI raised customer-support productivity nearly 14%, with the biggest gains among less-experienced workers.

That said, task-level efficiency still hasn’t produced a visible macroeconomic boom. Productivity rose only 1.4% annualized in Q2, behind its long-run 2.1% pace. Manufacturing productivity grew just 0.5% annually during the current cycle, underscoring that the gains remain concentrated in cognitive services.

Diffusion remains another constraint.

Census data show that just 17% to 20% of U.S. businesses used AI through early May 2026. A Danish study covering 25,000 workers found chatbots saved nearly 3% of time but had no significant effect on earnings or recorded hours.

On top of that, MIT economist Daron Acemoglu estimates that AI might raise total factor productivity by no more than just 0.66% over a decade, or even less than 0.53% if more difficult tasks are considered.

What Solomon’s outlook means for investors 

Solomon’s comments offer investors more reason to stay constructive, but it’s wise not to ignore valuation or execution.

The pillars he talks about can extend the cycle, but a lot of that optimism is priced into years of AI-powered growth. That said, it’s imperative to distinguish between the spending beneficiaries and productivity beneficiaries.

Chipmakers, data-center operators, and power suppliers are monetizing the buildout now. However, it remains critical for their customers to prove that their AI spending will translate into lower costs, higher growth, or wider margins. If those returns emerge, AI could effectively broaden out earnings beyond infrastructure leaders. 

Balance sheets also matter. Solomon feels that there’s little systemic credit risk because many of the largest borrowers generate a ton of cash flow. 

A useful example is Google parent Alphabet (GOOG) stock. Despite a massive $44.9 billion capex in Q2, primarily for AI infrastructure, it generated $185.7 billion in trailing 12-month operating cash flow. It also maintained $242.5 billion in cash and securities, along with $98.2 billion in long-term debt.

That could reduce near-term danger, but it doesn’t eliminate misallocation. Investors need to continue monitoring free cash flow after considering capital spending, debt growth, interest coverage, and whether AI sales can scale more quickly than depreciation and financing costs.

Perhaps the macro signal is productivity. If we see sustained output-per-hour growth over the pre-pandemic trend, it would support higher profits without reigniting inflation, potentially paving the way for strong growth alongside easier monetary policy.

The obvious conclusion is selective optimism. Favor companies already converting AI into measurable sales, margins or customer savings, while treating distant productivity promises cautiously. 

Related: Scott Bessent has surprising answer for U.S. debt fears

Ridley Scott has some less-than-kind words for “Alien: Romulus” as he returns to “help” the franchise with a new sequel.

A deal potentially worth $120 billion would normally be the kind of announcement investors celebrate.

For Marvell Technology (MRVL), it produced the opposite reaction.

Shares of the chip designer dropped more than 8% to $221.60 in early trading Aug. 28, putting the company on pace to lose more than $17.4 billion in market value.

The company performed well. Management raised its revenue forecast, according to Reuters. Analysts lifted price targets. Marvell has one of the biggest new artificial intelligence opportunities in the semiconductor industry in a custom-chip deal with Alphabet (GOOGL) subsidiary Google.

But Wall Street dumped the shares.

The reason provides a useful lesson for ordinary investors chasing the AI boom: A giant contract is not the same thing as giant revenue today.

The Google opportunity will be much larger in FY 2029, Marvell CEO Matt Murphy said in a post-earnings call. Marvell’s previous guidance through fiscal 2028 also included some of the Google-related revenue.

So investors who had expected the $120 billion headline to immediately translate into dramatically higher forecasts found that part of the payoff is years away.

And when a stock trades at nearly 59 times forward earnings, waiting becomes expensive.

Marvell’s $120B number has a timing problem

Marvell has emerged as one of Wall Street’s biggest beneficiaries of the rush to build custom AI chips.

Big tech companies are increasingly designing their own custom semiconductors to boost performance and cut the enormous costs of AI computing.

Reuters reported that Big Tech’s AI spending is expected to exceed $740 billion this year. That spending surge helped Marvell shares nearly triple in 2026 before the latest selloff. Those gains also altered investors’ expectations for the company.

Exceeding expectations simply wasn’t enough.

The new custom-chip arrangement for Marvell with Google could generate as much as $120 billion in revenue through fiscal 2033, Reuters reported. That sounds like a huge amount compared to Marvell’s current business.

Related: Buffett’s Berkshire is doubling down on Google

But Murphy told investors that the company’s existing custom revenue targets through fiscal 2028 already included some Google revenue.The much bigger contribution is expected to begin in fiscal 2029.

That was the key difference in the post-earnings debate.

Expectations were higher mostly because of the Google deal, Morgan Stanley analysts said, and its contribution was largely already priced into earlier company guidance. Thus, Wall Street wasn’t discovering a whole new $120 billion opportunity after earnings.

Investors were getting additional information about when a known opportunity would actually be reflected on the financial statements.

Marvell is growing fast, but Wall Street wanted faster

The selloff is even more striking considering that Marvell’s underlying growth outlook is not weak at all.

The company expects revenue growth of about 45% for fiscal 2027.

Revenue is projected to reach around $18 billion in fiscal 2028, aided by continued growth in its data-center business.

Those are big numbers for many semiconductor companies. But for Marvell, they ran counter to expectations that had grown even faster than the business itself.

That’s an increasingly important distinction for the AI business.

Investors aren’t asking if AI semiconductor companies will grow anymore. They are asking themselves whether those companies can grow faster than the assumptions already baked into their share prices.

More Tech:

Anthropic-powered AI model sends shocking message to employee

Rocket Lab clears 1st hurdle in its biggest satellite deal

Apple’s $54 billion iPhone machine may be about to break its biggest ritual

That is a particularly high bar given the valuation of Marvell.

Marvell was trading at 58.41 times its forward earnings. Competitor Broadcom (AVGO) trades at 32.15x. That means Marvell trades at an earnings multiple some 82% above Broadcom’s forward multiple.

Investors could justify such a premium if they believe Marvell’s future earnings will grow much faster.

But it also allows for less disappointment.

When a stock trades at nearly 60 times earnings, a good quarter can still be a bad day if the next wave of profits is later than investors expected.

Investors can learn something from Marvell’s plunge

The lesson here extends well beyond semiconductor stocks. Suppose a company announces a gigantic contract. The first number investors naturally focus on is the total potential value. With Marvell, that number is up to $120 billion.

But there are several other questions that matter just as much:

How many years will the revenue be spread across?

How much was already included in Wall Street forecasts?

When does the largest contribution actually begin?

What profit margins will the work generate?

How expensive is the stock before the new revenue arrives?

Marvell’s selloff is a case study in what happens when investors pay a lot of attention to the first number and then get less exciting answers to the rest. That’s a far cry from a $120 billion opportunity that extends through fiscal 2033 to $120 billion of incremental revenue coming in over the next several quarters.

And management has now said that the Google business doesn’t become meaningfully more important until fiscal 2029.

That doesn’t necessarily make the opportunity any less valuable for long term holders. But for a market that has nearly tripled Marvell’s stock this year, it alters the timing of the payout.

Analysts still see a far bigger earnings opportunity

What’s remarkable about the selloff is that Wall Street isn’t suddenly giving up on Marvell.

At least eight brokerages raised their price targets after the results, Reuters reported.

The median analyst target increased to $275. That was around a 13.8% gain from Marvell’s previous close.

Melius Research was also bullish on the longer-term opportunity, according to Reuters.

Marvell’s analysts said its custom-chip business, potential business with Microsoft, and further growth in AI connectivity could eventually result in a big expansion of earnings. They wrote, “$20 in EPS power before the end of the decade look[s] realistic.”

That figure offers another way for investors to see why Marvell merits such a lofty valuation.

The market is not necessarily valuing the company on what it makes today. It’s valuing Marvell at what its AI business could look like in a few years. That can lead to explosive gains when expectations rise. Even a slight change in the timeline can also result in violent sell-offs.

Marvell’s $120 billion promise is testing investor patience.Bloomberg / Getty Images

The AI boom has created an expectations problem

A bigger problem is surfacing in the semiconductor industry.

Hundreds of billions of dollars are being invested in AI infrastructure by the big tech companies.

Chip designers relying on that spending are posting huge growth rates. And investors have rewarded some of those businesses with fantastic valuations. But eventually those stocks run into a mathematical problem. The better the story, the more future success is baked into the shares.

Marvell’s stock has more than tripled before falling Friday, according to Reuters. Revenue is expected to increase by 45% in fiscal 2027. It pulls in roughly $18 billion in revenue in fiscal 2028. It has a $120 billion Google opportunity through fiscal 2033.

Those aren’t the numbers of a company that has lost its AI chance. Still, the stock plunged more than 8%. That raises serious concerns for investors about the state of the AI trade. Wall Street increasingly wants growth, and growth sooner than expected.

Marvell’s Google deal may still pay off, just not fast enough for everyone

So it’s the fiscal 2029 timetable that investors might want to pay closer attention to than the $120 billion headline.

If Google’s contribution lives up to Marvell’s expectations, the company may eventually have an excuse to explain much of the optimism already priced into the stock.

Melius Research thinks the Google relationship, Microsoft prospects, and connectivity business could produce roughly $20 in earnings per share by the end of the decade.

This wave of price-target increases shows analysts remain broadly constructive.

But Marvell’s selloff reveals the flipside of owning an expensive AI stock.

A company can win a giant customer. It can raise its forecasts. It can project 45% annual revenue growth.

And Wall Street can still wipe out more than $17 billion of its market value because investors had expected the good news sooner.

Perhaps that’s the most valuable number in Marvell’s earnings story for the average investor.

The company didn’t lose its Google opportunity in a day.

Investors just found out some of the $120 billion future they thought they were buying is further away than they thought.

Related: Ackman just walked away from Google

Dua Lipa’s “Training Season” returns to three U.K. charts after going viral on TikTok and the release of the new ‘Training Season’s Over’ EP.

Perplexity today launched hybrid compute for its agentic platform, Computer, a system that lets a single AI agent split its work between frontier models running in the cloud and smaller open-weight models running locally on Apple silicon Macs — routing sensitive data to the local machine so it never leaves the device.The company says it is the first time an AI agent can begin a task in the cloud and dynamically hand off the confidential portions of that same task to a model running on the user’s own hardware, without restarting the job or losing context. The feature becomes available today through Perplexity’s desktop app for enterprise customers that opt in, as well as Pro and Max subscribers, on any Apple silicon Mac running macOS 15 or later.”Hybrid is really compelling because it’s often the work that requires confidentiality that is the most important to get right, and so the accuracy really, really matters,” Jon Staff, who leads Perplexity’s macOS and iOS engineering teams, said during a press briefing attended by VentureBeat. “By combining these two together, we can get that maximum intelligence from the frontier models, but we also get the security and the privacy that comes with local.”How Perplexity’s on-device privacy gate keeps sensitive data off the cloudThe architecture works like a dispatcher. A frontier model in the cloud breaks a task into subtasks and routes each one to the appropriate place. Web research, long-horizon planning and heavy reasoning run in the cloud, while anything touching private files, local data or actions on the device gets delegated down to a subagent running on the Mac itself.The linchpin is what Perplexity calls a Privacy Gate: a company-trained classifier that runs on the device and scans for personally identifiable information — names, addresses, account numbers, secrets — before anything is transmitted to the cloud. When the gate flags sensitive content, the user chooses whether that portion of the task runs locally or gets shared.”What we wanted to do is make sure anything that’s shared to that cloud orchestrator is safe,” Staff said. “We built and trained our own PII classifier that integrates directly into the Mac app.”He described the handoff in detail: “The cloud orchestration will break down the task based on the prompt and figure out how to route it to different subagents… it’s going to delegate that down to a sub-agent running on your Mac, and then that portion of the task is run entirely local. None of those tokens go to the cloud.”The economics matter, too, for a company that meters cloud usage through credits. Tokens generated locally cost nothing. “You’re paying for the electricity, you’re paying for the hardware, so we’re not charging you for that,” Staff said. “The only thing the credits are used for is the orchestration and the delegation.”Lawyers, private equity firms and a founder in an Uber: hybrid compute in actionPerplexity built its demonstrations around exactly the kind of work most professionals would never hand to a cloud-only agent. In the first, a lawyer on deadline updated a draft brief against privileged case files stored on a Mac while a cloud agent simultaneously pulled public case law from the open web — sending out, Perplexity says, only anonymized legal questions. “At no point did their privileged information get shared to the cloud,” Staff said. “It never left the Mac.”In the second demo, a private equity associate’s agent reworked a financial model against confidential management projections, benchmarked the deal against public comparables and produced a fifth iteration of an investment committee deck. The task ran roughly 40 minutes in the background with no human input — work that would have taken hours of manual stitching between local spreadsheets and cloud research.The third demo emphasized continuity across devices. The founder of a pottery shop, riding in the back of an Uber, kicked off a marketing analysis from her iPhone. Computer asked permission to reach her Mac at the studio, fired up the local subagent to process her customer interviews and revenue data, and combined that with cloud research on competitors’ public pricing. “It doesn’t matter how far away she is from her computer,” Staff said.”Tasks like this aren’t possible in a fully local or a fully cloud setup,” he added. “You need that security of the local and the privacy, but you also need the intelligence of the frontier.”Why a Chinese-made Qwen model on enterprise Macs is raising eyebrowsThe launch model lineup immediately raised a pointed question. At launch, users can choose among three local models: Google’s Gemma E4B, Alibaba’s Qwen3.6 35B-A3B, and a Perplexity post-trained version of Qwen3.6 35B — the company’s recommended option. Asked by VentureBeat whether enterprise or government customers had raised concerns about giving a Chinese-developed model access to their machines, Staff argued that local inference neutralizes the geopolitical risk.”The great thing about these models is that they are open weight. We’re able to evaluate them ourselves,” he said. “When that model is running locally on your computer, the data is not going outside of your computer itself… You’re not actually sending those tokens to some cloud provider that’s hosted in another country. In fact, all of Perplexity’s models are U.S. hosted.”He added that macOS’s built-in sandboxing framework, known as Seatbelt, constrains what the agent can actually do on a machine: “If local execution is trying to do something that it shouldn’t, it’ll just point blank stop it and it’ll request permission from the user.” Perplexity does not currently allow unrestricted “YOLO mode” execution, he said, though “I wouldn’t be surprised at some point if we allow certain people to do this.”For enterprises, admins can set a single organization-wide sensitivity policy and audit a full record of what leaves each device — a feature aimed squarely at compliance teams in law, finance and healthcare. Questions remain on the consumer side, however. Pressed on how usage data feeds model training, Staff pointed to Perplexity’s incognito mode and a long-standing opt-out toggle, and said enterprise contracts can include zero-data-retention terms. A company spokesperson said Perplexity is “not using it for post training” globally and promised to follow up with specifics on non-enterprise accounts.The enterprise privacy problem hybrid AI is trying to solveThe announcement lands amid a broader industry reckoning with a stubborn problem: the most valuable enterprise work involves exactly the data companies are least willing to send to someone else’s servers. NIST’s generative AI risk profile flags data privacy and information leakage among the technology’s central risks, and McKinsey’s research on the state of AI has consistently found that organizations struggle to move from experimentation to value capture, with data governance among the chief obstacles. Gartner, for its part, named hybrid computing among its top strategic technology trends for 2025, anticipating architectures that blend compute across environments.Perplexity is betting that the answer is not choosing between cloud intelligence and local privacy, but building the orchestration layer that arbitrates between them in real time. It is a defensible position for a company that has always styled itself as a neutral broker — “Perplexity is like Switzerland in that we work with everyone,” a company representative said at the briefing — sitting at the application layer above whichever models happen to lead at any given moment.”Anytime one of these gets better, Perplexity gets better,” Staff said of the interplay among local models, frontier models and Apple’s chips. “That’s the really cool nature of where we sit in this application layer, orchestrating all the different pieces together.”From $520 million startup to $20 billion agent platform in three yearsHybrid compute caps an extraordinarily aggressive product run. Perplexity launched its Comet AI browser in July 2025, initially for $200-a-month Max subscribers — an early bid to make agents, not chat, the interface to computing. Computer, its full agentic platform, arrived in March 2026, followed by desktop apps for Mac and Windows. Just last week, the company launched a local-first version of Computer on NVIDIA’s DGX Spark hardware, which starts on the user’s device and escalates to cloud models only with permission. Today’s launch inverts that flow: cloud-first, delegating down.The business trajectory has been equally steep. Perplexity was valued at $520 million in January 2024; by September 2025, the company had finalized a funding round at a $20 billion valuation. Along the way it made an audacious $34.5 billion bid for Google’s Chrome browser during Google’s antitrust remedies fight, and Bloomberg reported that Apple executives held internal talks about acquiring the company — a striking backdrop for a product now built to showcase Apple silicon.The strategy is not without headwinds. Reuters reported in July that Reddit’s data-scraping lawsuit against Perplexity survived a motion to dismiss, part of a wave of copyright and data litigation facing the company — context that makes its privacy-forward positioning both commercially savvy and reputationally necessary. And practical constraints remain: Perplexity recommends at least 32GB of unified memory for the better tier of local models, Staff was candid that the smallest option “significantly underperforms” the larger Qwen models, and Windows and Linux support will come only later.The deeper question is one users cannot easily inspect. The Privacy Gate is itself a machine learning classifier, and classifiers miss things; a false negative means sensitive data reaches the cloud anyway. Perplexity’s answer is transparency — users can expand and review exactly what the gate flagged before anything is sent, and enterprises get device-level audit logs. But the pitch, at bottom, asks professionals to trust one AI to decide what another AI is allowed to see. For an industry that has spent three years telling lawyers, bankers and doctors to keep their most sensitive work away from the cloud, Perplexity’s wager is that the fix was never to build a higher wall — it was to build a smarter gate.

Instead of merging with Delta Airlines, CEO Scott Kirby should direct his energies to operating United Airlines like it.