SUCCESS

New research has revealed that an upcoming real estate development is expected to be home to the world’s closest residential complex to a Disney theme park.

Indie K-pop label hello82 revealed the lineup for 82WORLD: LA, its first outdoor festival, running Oct. 17-18 at Los Angeles State Historic Park.

Venture capital has an old rule about timing. You raise money when investors are eager, not when you need it, because the gap between those two moments can slam shut without much warning.For most of startup history, the window opened every 18 to 24 months. Founders closed a round, spent a year or two building, then came back with numbers that justified a higher price. The rhythm gave everyone time to breathe, and it gave boards time to ask hard questions about whether the growth was real.Artificial intelligence has shredded that calendar. Rounds that once took months to assemble now close in weeks, and the hottest companies field offers they never solicited. Investors who watched OpenAI and Anthropic compound in value are determined not to miss the next name on the list, so the checks now arrive before anyone asks for them.Even by those standards, one company has compressed the cycle to a length for which I struggle to find precedent. Cognition AI, the startup behind the autonomous coding agent Devin, is in early talks for a funding round that would value it at $40 billion or more, according to Bloomberg. Its last raise closed less than three months ago.Why AI coding became venture capital’s favorite betSoftware development is the first profession the artificial intelligence industry has tried to automate at industrial scale. The people building the models understand engineering work intimately, and code offers something rare in AI: an output you can actually test. It either runs or it does not, so corporate buyers can measure the payoff and open budgets quickly.More Artificial Intelligence:Nvidia’s biggest risk may be coming from inside its customer baseMark Zuckerberg sends controversial message on the future of jobsMichael Burry delivers brutal verdict on $420 billion Al giantThat measurability has turned coding tools into a gold rush. Cursor maker Anysphere reached $2 billion in annualized revenue earlier this year and discussed raising money at a $50 billion valuation in March, Bloomberg confirmed. SpaceX then agreed to acquire Cursor outright, in a deal expected to close as soon as this week, the outlet reported.A quick translation before the numbers get bigger. An annualized revenue run rate takes the most recent stretch of sales and multiplies it across a full year. It flatters fast growers, which is exactly why startups love citing it, and why the figure deserves a raised eyebrow alongside the applause.The big labs want the same territory. Anthropic, OpenAI, and SpaceX have all dedicated large chunks of their businesses to engineering products, a crowd that has made coding what Bloomberg called “one of the industry’s hottest spheres.”Cognition has played offense in that consolidation, too. The company bought rival Windsurf in July 2025, days after Alphabet’s (GOOGL) Google unit hired away Windsurf’s CEO and research leads, and after merger talks between Windsurf and OpenAI fell apart, according to Benzinga.When this many well-funded players fight over one market, private valuations stop behaving politely. That is the backdrop for this week’s news.

Cognition, maker of Devin, may raise $1B at 40x revenue three months after last round.PuiStocker65 / Getty Images

Cognition’s $40 billion valuation talks arrive at record speedCognition is in early discussions with investors to raise more than $1 billion at a valuation of at least $40 billion, up more than 50% from the $26 billion price it set in May, according to Bloomberg. Prospective backers approached the company rather than the other way around, and the talks could still change terms or fall apart entirely, the outlet’s sources cautioned.Related: SoFi Makes Major AI Investing MoveThe revenue trajectory explains the appetite. Cognition’s annualized run rate is now approaching $1 billion, roughly double the figure it showed during the May financing, Bloomberg reported. Doubling sales in about one quarter is the kind of growth that makes spreadsheet models blush. It is also the only kind that can justify asking for a 50% markup roughly 90 days after your last one.The May numbers were already striking on their own. CEO Scott Wu confirmed a $492 million annualized revenue run rate at the time of that round, with enterprise customers expanding their use of Devin by 50% month over month for six straight months, according to TechCrunch.Devin is not pitched as a replacement for human engineers. The agent takes on “long-tail grunt-work that many programmers dislike,” such as modernizing old software, according to TechCrunch. Customers include Citigroup (C) and Goldman Sachs (GS), according to PYMNTS.The resulting valuation climb reads like a misprint. Here is what investors have paid at each step:April 2024: About $2 billion, The Wall Street Journal reported.September 2025: $10.2 billion after a $400 million round led by Founders Fund, TechCrunch confirmed.May 2026: $26 billion after a $1 billion raise, according to Bloomberg.August 2026: Early talks at $40 billion or more, based on the same Bloomberg report.Founded in 2023 by Wu, Steven Hao, and Walden Yan, the company has multiplied its price tag roughly 20 times in about 28 months. Most startups take a decade to travel that distance, and most never do.What a 40 times revenue multiple means for your portfolioI ran the arithmetic, and it is bracing. A $40 billion valuation set against a run rate approaching $1 billion works out to roughly 40 times annual revenue. Cursor’s March talks, for comparison, implied about 25 times its $2 billion run rate, based on the figures Bloomberg cited.Retail investors cannot buy Cognition shares, but the number still matters for anyone holding public AI names. Private markets are setting the reference prices that public companies get measured against, and a 40 times multiple only makes sense if revenue keeps doubling on a quarterly clock.My analysis of past funding manias keeps landing on the same lesson. Velocity cuts in both directions. If Cognition keeps doubling, today’s price will eventually look as quaint as its $2 billion valuation does now. If growth merely slows to very good, the repricing will ripple through every AI position in your account, because these private marks anchor the whole sector’s story.If you write software for a living, the same chart reads differently. A company approaching $1 billion in revenue by selling automated grunt work is a signal about where entry-level engineering tasks are heading, and it is worth taking seriously, even if you doubt the price tag.None of this requires you to believe the hype or dismiss it. It requires you to notice that the smartest money in the world is repricing software labor in real time, and your retirement account, whether you like it or not, is along for the ride.Watch two things from here. Whether this round actually closes at $40 billion or better, and whether SpaceX completes its Cursor purchase on schedule. Together, those answers will tell you whether the money chasing AI coding is still accelerating or finally pausing for breath.Related: Anthropic just made a move that changes the AI investing story

Every gold rush produces two kinds of fortunes. There are the people who find the gold, and there are the people who sell them the shovels. History remembers the first group and quietly enriches the second.The artificial intelligence (AI) buildout has followed that script almost to the letter. Investors who missed Nvidia (NVDA) crowded into what Wall Street calls the picks and shovels trade, buying the companies that deliver power and cooling to data centers rather than the chips humming inside them.That trade worked. Then it got crowded, and crowded trades punish disappointment. Vertiv Holdings (VRT), the largest pure play in data center thermal management, trades well below the all-time closing high of $376.15 it set on May 14, according to Macrotrends.So the money has moved again, one layer further down the stack. Investors hunting the next round of AI winners are now buying the companies that make vacuum pumps, heat exchangers and ultra high purity gases, the deeply unglamorous inputs without which a chip fab is an empty building.

AI’s next winners make vacuum pumps, not chips.Rasi Bhadramani / Getty Images

Why chip fabs cannot run without pumps and gasesA modern semiconductor fab is essentially a very expensive vacuum. Etching and deposition steps happen in chambers pumped down to near space conditions, and the exhaust from those chambers is toxic enough to require its own abatement equipment.Feeding those chambers takes nitrogen, argon, helium and a long list of specialty gases delivered at purity levels measured in parts per billion.None of that shows up in a Nvidia keynote. All of it shows up in the capital budget of every fab under construction.More AI Stocks:Bank of America sends blunt message to Nvidia stock investorsGoldman Sachs just answered the biggest question about AlASML’s recent dip gives rare opening for Al investorsThe scale is set upstream. ASML said its order intake stayed very strong through the first half and that it is planning roughly 30% more low numerical aperture extreme ultraviolet capacity for 2027, according to a company statement.Every one of those lithography tools lands inside a fab that needs pumps, abatement and gas lines wrapped around it.The same logic runs through the data center itself. Cooling is the piece investors already understand, and my colleagues have covered how hyperscalers are racing to lock in chiller capacity years ahead of need.What gets missed is that the fab side of the equation is a consumables business.Pumps need service. Gases get consumed and reordered. Abatement systems require replacement parts on a schedule for as long as the plant is running.Related: Overlooked AI stock is growing faster than NvidiaCooling equipment, by contrast, largely gets sold once per build.European makers of vacuum pumps, heat exchangers and specialty gases have become “lesser-known winners” of the AI boom, according to Bloomberg.The order books tell a very different AI storyAtlas Copco (ATLCY), the Swedish industrial group whose Vacuum Technique division supplies pumps and exhaust management systems to chipmakers, posted organic order growth of 59% in that division in the second quarter.Group orders hit a record 50.95 billion Swedish kronor. Demand “improved significantly, particularly from the semiconductor industry,” said Chief Executive Vagner Rego, according to a company statement.When I ran the divisional numbers against the group total, one thing stood out.Vacuum Technique represents roughly a quarter of Atlas Copco’s order intake but produced the overwhelming share of growth. The rest of the portfolio grew at ordinary industrial rates in the low teens.Strip the semiconductor exposure out and this is a solid European machinery company. Leave it in and it is an AI stock that nobody files under AI.Management is not treating this as a cycle to harvest. Chief Financial Officer Peter Kinnart said demand in the vacuum business is “very extraordinary,” and that some factories need to more than double output, according to Investing.com.Linde (LIN) tells a parallel story on the gas side. Electronics sales rose 18% year over year in the second quarter, and the company’s contracted sale of gas backlog reached a record after a $1 billion U.S. contract to supply advanced node fabs.“The electronics pipeline is looking healthy,” said Chief Executive Sanjiv Lamba, according to TradingView.Here is the scoreboard for the quarter:Atlas Copco’s Vacuum Technique orders grew 59% organically, according to Atlas CopcoLinde’s sale of gas backlog hit a record $8.1 billion, according to LindeLinde’s total project backlog reached $11 billion, according to LindeASML (ASML) plans to add 30% to its 2026 low NA EUV capacity for 2027, according to ASMLVertiv grew second quarter sales 24% and raised guidance, yet shares fell, according to MorningstarWhat owning the boring layer actually costs youThe case for these names is not that they are cheap. It is that they sell into fab construction schedules locked in years ahead, and then keep selling parts and gas for the life of the plant.That is a different revenue shape from selling a rack of cooling equipment once.My analysis of the two business models comes down to duration.A cooling order is a project with a delivery date. A gas contract is closer to an annuity with a fab attached, and Linde’s backlog represents contracted revenue on plants that have not been built yet.That is why I weigh backlog quality more heavily than headline growth in this corner of the market. Growth tells you what just happened. Backlog tells you what has already been signed.Now the honest part. Atlas Copco’s vacuum business shrank last year before this year’s surge, so the cyclicality is real, not theoretical.Most of these companies also trade in Stockholm, Paris and Zurich, reaching U.S. investors through American depositary receipts with thinner volume and currency exposure attached. Your S&P 500index fund owns almost none of this.Domestic investors do have entry points. Linde lists on the Nasdaq, and Ingersoll Rand (IR) sells vacuum and gas handling systems out of Davidson, N.C.Entegris (ENTG) supplies the materials and filtration that keep fab chemistry clean.None of the three is a pure play on this theme, which cuts both ways. You get diversification and you get dilution of the exact exposure you were shopping for.Fidelity’s managers have been building positions in the electrical and mechanical layer of this buildout, as TheStreet reported in December.The trade that survives an AI spending pauseAsk what happens if AI model demand disappoints. Chip design cycles turn fast, and Intel’s cheaper inference push shows how quickly the architecture argument can shift.Fabs do not turn fast. A plant approved in 2026 gets built through 2028 and runs for two decades, consuming pumps and gases the entire time.That is the quiet argument for owning the layer below the layer everyone already found. It is not a bet on which model wins. It is a bet that the buildings keep running.The names are hard to pronounce and harder to explain at a dinner party. Their order books do not care.Related: AI agents are quietly rewriting how the internet works

The conservative news network posted $2.9 million in net income in its latest quarter.

The original partnership was set to last through 2027, but now the two high-energy global brands signed a deal to extend the partnership through the 2028 season. 

The SEC was ready to release at least some part of the innovation exemption alongside its now-canceled open meeting for “Reg Crypto” on Friday.

Betting against AI still looks like a dangerous game.In particular, one fast-growing AI neocloud stock has surged more than 200% in 2026, as investors continue to pour money into companies facilitating the AI boom. Demand looked relentless, and the market’s been pricing in plenty more growth, even though investors appear to be in “show-me” mode with regard to AI.”Big Short” Michael Burry, though, is betting something has gone wrong.He continues to lean further against AI, adding to his bearish positions involving Micron (MU), Oracle (ORCL), and the iShares Semiconductor ETF (SOXX), while increasing another short against one of AI’s hottest stocks.Why is Michael Burry shorting Nebius after earnings? Nebius’ (NBIS) earnings revealed a booming business, but Burry thinks there’s trouble beneath the headline growth.“Nebius is what the top of a boom looks like,” Burry wrote in a Substack post after adding to his short position at $247.For perspective, Nebius has been a tremendous wealth compounder over the past year and essentially functions as a neocloud.More Manager Buy/Sells:Michael Burry increases his bet against popular chip giantWarren Buffett reveals he broke his own investing patternMark Cuban bets on MLB with Athletics minority stakeIt’s basically an AI-focused cloud provider building potent data centers packed with pricey GPUs, renting the compute power to businesses training and running AI models.Burry’s criticism centers on the pricing of Nebius’ compute contracts. Its management said it could potentially sell all of its 2027 capacity through one- to three-year agreements but is deliberately keeping some supply available for shorter-term customers.Those short-duration deals could command nearly $40 million to $50 million per megawatt, sometimes more. Conversely, midterm contracts are priced around $20 million to $25 million per megawatt.That’s the part Burry finds unusual.Typically, longer commitments provide a supplier with stronger visibility and could justify more attractive economics. Here, customers are looking to pay nearly twice as much for immediate access to compute. One way to look at it is from a scarcity lens, but Burry feels customers are attaching dramatically more value to AI capacity today than to the same capacity further out. In the commodities space, that relationship is often characterized as “backwardation.”However, for AI infrastructure, Burry’s question is essentially: What is the market expecting to lose value so quickly?“Power does not depreciate,” he wrote.Moreover, Burry questioned Nebius on extending the depreciation life of its servers from four years to five. He wonders why Nebius is assuming its hardware will remain economically useful longer, even as contract pricing is expected to imply a sharp drop in the value of future compute.Nevertheless, Burry’s short is essentially a bet that investors expect today’s AI boom to last a lot longer than the underlying assets could realistically support.

Michael Burry increased his Nebius short after questioning the company’s AI economics.Astrid Stawiarz/Getty Images

What stood out in Nebius’ earnings?Revenue exploded 454% year over year to $582.3 million, while core Nebius AI revenue jumped 514% to $575 million, indicating that the company’s growth is now spearheaded by AI infrastructure demand.Annualized run-rate revenue reached $3 billion, up 598% from a year earlier and 56% from March, an unusually steep acceleration in compute demand.Profitability improved dramatically. Adjusted EBITDA swung to $236 million from a $21 million loss, while the margin climbed to 41%. Nebius AI itself generated a remarkably robust 50% EBITDA margin.Operating cash flow surged to $2.3 billion, giving Nebius stronger internally generated liquidity as it looks to scale up quickly.Nevertheless, the growth is highly capital-intensive, with quarterly capex of roughly $5.7 billion, nearly 10 times reported revenue.
Source: Nebius Q2 Investor Relations Update
How has Nebius stock performed? Nebius has clearly been one of the biggest winners of the AI infrastructure trade.According to Seeking Alpha data, the stock surged almost 210% year to date and 244% over the past year, punishing the S&P 500’s respective 13.2% and 20.2% gains. Even over six months, Nebius is up nearly 189%.Consequently, that tremendous momentum has driven its valuation expectations sky-high.Nebius stock trades at 73.1 times trailing sales and 20.5 times forward sales, compared to the sector median of just 3.7 and 3.4 times, according to Seeking Alpha.Perhaps the more interesting number is the forward price-to-cash-flow of 8.7 times (below the sector’s 20.7 times), which suggests investors are pricing in a dramatic cash-flow ramp. What makes Michael Burry’s investing strategy different? Michael Burry’s contrarian investing style often involves betting against red-hot stocks. His approach involves working backward, identifying gaps between investor euphoria and what the company’s economics can actually support.AI is the obvious example at this point.In Scion Asset Management’s final reported 13F, Burry held put options tied to 5 million Palantir shares and 1 million Nvidia shares. These shares attracted an underlying value of nearly $912 million and $187 million, respectively; those figures weren’t the amounts Burry actually spent on the options.That’s classic Burry, where he isn’t afraid of constructing asymmetric trades. Options allow him to strongly express his bearish view while defining the capital committed to the trade.Of late, that same skepticism has deepened into the AI infrastructure boom. Business Insider reports that Burry disclosed he was shorting Oracle at roughly $145, citing concerns about the tech giant’s ballooning debt load, long-term commitments, and off-balance-sheet obligations.While the bulls are still enamored with the flashy headline figures, Burry is asking what those lofty sales cost to produce, how much capex is consistently reinvested, and who ultimately bears the risk.“The fish have gotten very fat, very large, and easy to shoot,” Burry wrote while discussing AI companies and their liabilities.And on Aug. 12, Burry recommended rewatching “The Smartest Guys in the Room,” the documentary about Enron, writing that he feels “history is repeating.”However, he isn’t automatically bearish when everyone else is bullish.Perhaps the best example is his June purchase of MercadoLibre stock at nearly $1,630, after it had dropped 39% from its June 2025 high, underscoring his disciplined buying approach. He’s less interested in catching falling knives than in whether the market enthusiasm has moved too far away from the underlying economics. Related: Nvidia’s biggest risk may be coming from inside its customer base

A once-fast-growing beverage brand is making a major retreat from the physical spaces that helped build its identity and exiting an entire part of its business.After years of expansion, the company is closing its final brick-and-mortar locations and has already shut down the facility where it once produced its beverages. The moves mark a significant shift for a brand that built its reputation on an in-house manufacturing approach and a growing physical presence.Founded in 2018 in San Diego, California, JuneShine started as a homemade hard kombucha brand before expanding into canned cocktails.JuneShine is closing its last tasting roomsJuneShine built its brand on the concept of “honest alcohol for a healthier planet,” targeting health-conscious consumers by offering clean, environmentally sustainable ingredients.The company will close its two final company-owned tasting rooms on Aug. 28, ending the company’s brick-and-mortar operations.The affected locations include:JuneShine Scripps Ranch: 10051 Old Grove Road, San Diego, CaliforniaJuneShineSanta Monica: 2914 Main Street, Santa Monica, California”This was certainly not a decision that came easy, but the right one for what’s next,” JuneShine wrote in an Instagram post.The company has not shared plans to open another physical location. However, JuneShine said its products will remain available online and through retailers, including Walmart, Target, Whole Foods Market, and Total Wine & More, according to its website.”Rest assured, JuneShine isn’t going anywhere. You will still find us on shelves, at your favorite watering holes, and out in the real world,” the company added.The closures, therefore, mark a change in how JuneShine operates, rather than an end to the brand itself.JuneShine has already reduced its physical footprintThe decision to close the final tasting rooms follows several changes to JuneShine’s operations over the past several years.In 2019, JuneShine acquired the former 30,000-square-foot Ballast Point brewery in Scripps Ranch, turning it into a flagship brewery and taproom. The company invested $24 million in the project.In 2020, JuneShine confirmed plans to relocate its San Diego tasting room to a new 2,000-square-foot space in North Park at The Jackson on 30th Street. The tasting room ultimately did not open.Two years later, JuneShine expanded beyond California by opening a taproom in Brooklyn’s Williamsburg neighborhood. That location closed in 2024.The company’s manufacturing footprint then underwent a more significant change in March 2026, when JuneShine ceased in-house brewing, shut down its brewery, listed the facility for sale, and eliminated 24 jobs.The move shifted production to third-party manufacturers as JuneShine sought to improve efficiency and profitability and focus more heavily on product development and brand expansion, SanDiegoVille reported.Now, with its final two tasting rooms scheduled to close, JuneShine will no longer operate company-owned physical locations.

JuneShine closes its final two physical locations.Illustration by Kira Hofmann/Photothek via Getty Images

Why JuneShine is moving away from physical locationsJuneShine has not publicly provided a detailed explanation for why it is closing its final tasting rooms. However, the move comes after a broader restructuring of its physical operations.The company has already moved away from owning and operating its own brewery, instead relying on third-party manufacturers to produce its beverages. Closing its tasting rooms could further reduce the fixed costs and operational responsibilities associated with maintaining physical facilities.Here’s some of my previous coverage of closures:Grocery giant rethinks supply chain plans as store closures mountPopular beverage chain closing multiple locations nationwide17-year-old Mexican restaurant chain closes all locationsFor smaller beverage companies, owning production facilities can require significant investments in equipment, labor, maintenance, and real estate. Outsourcing some or all production can allow a business to shift those responsibilities to specialized manufacturers, although the financial impact can vary depending on a company’s scale and manufacturing arrangements.Operating a taproom or brewery can also come with significant real estate and build-out costs. Specialized equipment and building requirements can add to the expense of opening and operating these businesses, according to Wooden Hill Brewing Company. JuneShine’s recent moves suggest the company is placing greater emphasis on its beverage products and retail distribution rather than maintaining company-owned facilities.That shift could give JuneShine greater flexibility to concentrate its resources on product development, distribution, and brand expansion while relying on outside partners for manufacturing and retail distribution.For consumers, however, the change means JuneShine’s remaining tasting rooms will soon disappear. The brand itself will continue through its online store, retail partners, and other locations where its beverages are sold.Related: Popular beverage chain closing multiple locations nationwide

After past snubs, social-media site Reddit has landed a spot on the S&P 500, sending shares on an after-hours rally Thursday.