Newsletter · · Ashutosh Agarwal

Software Stocks Rally as SaaSpocalypse Fears Fade for Now - Weekly SaaS / Software Podcast Recap - Week of August 16, 2026

Weekly SaaS and software podcast roundup for the week of August 9-16, 2026. Software stocks rallied off their lows as the SaaSpocalypse fear faded, even as investors argued that the whole AI and software complex now rests on OpenAI and Anthropic and its reported $2 trillion IPO.

Weekly SaaS / Software Podcast Recap

Week of August 16, 2026: Software Stocks Rally as SaaSpocalypse Fears Fade for Now

What the podcasts were saying about software this week, covering the last 7 days (Aug 9–16, 2026).

Top of mind

Two big storylines ran through almost every software conversation this week.

The first: the "SaaSpocalypse" fear that gripped the sector earlier this year has, for now, gone on vacation. Software stocks have rallied hard off their lows, and several hosts and operators openly declared the "software is dead" call premature. But nobody thinks the underlying pressure has vanished, the debate has simply moved from "will SaaS die?" to "how does software get paid when its own product destroys the thing it used to bill for?"

The second: the entire AI and software complex increasingly rests on two private, money-losing companies, OpenAI and Anthropic. With Anthropic reportedly racing to a roughly $2 trillion October IPO, investors spent the week arguing over whether that number is genius or madness, and whether a wobble at either lab would send the whole chain into reverse.

Underneath those two, there was a rich run of specific company calls: Palantir (three different stances, from "best company in the world" to "worth under a dollar"), Adobe as a possible value trap, Workday getting take-private interest, Salesforce suffering its first visible big-enterprise pullback, Shopify's AI bragging, Cloudflare's wild agent-traffic forecast, Canva's growth scare, and Databricks' fresh $190 billion round.

1. Dominant Themes

Theme 1: The SaaSpocalypse Relief Rally, the Funeral Was Cancelled

"SaaSpocalypse" is the nickname for the sell-off earlier in 2026 when investors feared AI would make traditional software companies obsolete (the idea that anyone could just "vibe code" (i.e., build software by describing it to an AI) their own version of Salesforce over a weekend).

  • On TBPN (Aug 14), the hosts took a victory lap for calling the panic overdone back in February: "Two trillion dollars of market cap lost across the SaaSpocalypse... But a lot of it's come back." Their point: the beaten-up names turned out to have "sources of strength" that don't fit the "just a big pile of code" story: network effects, distribution, and taking only a tiny slice of a customer's revenue. Their favorite example is Shopify: "A business that I know very intimately... Shopify Plus... is around $1,000 a month in cost and the business is doing almost a hundred million a year. And so you're at 0.1%", too small a bill to bother replacing. Cybersecurity emerged as a clear AI winner: Palo Alto Networks up 121% over the year to a ~$320B company, CrowdStrike up 107% to ~$230B.
  • Freshworks Chief Product Officer Srini Raghavan, on DataFramed (Aug 10), put it memorably: "The SaaSpocalypse as it was predicted was a wrong prediction, but it was pointing... to the right patient." He reframed it as "SaaS consolidation", buyers revolting against complexity, not against software. His striking stat: mid-market companies "lose about 25% of the money that they spend on SaaS... for integration" before they see any outcome. Winners will be platforms that turn that friction into results.
  • Even the bears conceded the mood shift. On The Real Eisman Playbook (Aug 14), Steve Eisman noted "the SaaSpocalypse narrative seems to have taken a bit of a vacation as software stocks have all rallied off the bottom."

Theme 2: The Pricing Reckoning, When Your Software Works Too Well You Lose Money (the Sizeocalypse)

This was the most intellectually meaty theme of the week. Traditional software is sold by the "seat" (per employee) or per transaction. If an AI agent lets a company do the same work with half the staff, the software vendor loses half its seats, so the better the product, the less it earns.

  • The Elon Musk Podcast (Aug 10) walked through analyst Gary Schope's framing of this as the "sizeocalypse": "The better your product works, the less money you make... the vendor is actively penalized for delivering a superior, highly efficient product because it destroys the seat count that their billing model relies on." The hosts cited HFS Research data that 62% of Global 2000 enterprises are actively renegotiating their SaaS contracts: "you are not looking at a bad quarter. You are looking at the structural collapse of an economic model." The proposed fix is the "outcome platform" model, charging for results, not access. Their example: digital-health firm Lantern took a claims process that "took over two weeks... reduced it to... about a minute." The catch, honestly acknowledged: proving exactly how much money you saved a client, and getting a procurement team to sign a deal based on a cut of their savings, is "a really heavy lift."
  • The cost to build software is collapsing toward zero, which is exactly why data and workflow ownership now matter more. Raghavan again: "You can vibe code any piece of software over a weekend, but the vibe coded software doesn't necessarily hit the mark... I can build a CRM over a weekend. I cannot actually get the data and the data sanctity."

Theme 3: Application Software vs. Infrastructure Software, Who Becomes a Dumb Data Pipe?

  • The standout analytical segment of the week came on Excess Returns (Aug 11), where a T. Rowe Price technology manager laid out why he thinks traditional application software is "in trouble" while infrastructure software is safer. His logic: systems of record (Salesforce, ServiceNow, Workday) risk becoming "a dumb data pipe into OpenAI or Anthropic" as the interface migrates to chat and agents: "this whole push by Salesforce to go headless is a concept that they know the direction of travel is that the user interface is going to migrate away from salesforce.com." He's "not seen a single compelling AI version of traditional enterprise software from a traditional enterprise software company." By contrast he's more comfortable with data-gravity and consumption businesses: SAP's ERP data, or "consumption software like Snowflake, dump all your data... into Snowflake and then let the models run on top."

Theme 4: The Great CapEx Test and Circular Financing

  • The week's biggest infrastructure news was NVIDIA lining up over $500 billion of financing with six giants (Apollo, Blackstone, BlackRock, Brookfield, Goldman Sachs, and KKR) treating GPUs as a borrow-against asset class. On The Rundown (Aug 11), host Zaid Admani relayed that BlackRock's Larry Fink "compared this to the creation of mortgage-backed securities in the 1970s," while noting the obvious discomfort: "mortgage-backed securities eventually became a pretty important character in the 2008 financial crisis." NVIDIA may backstop up to 25% of individual loans, feeding fears of "circular financing" (NVIDIA helps finance customers who then buy NVIDIA chips).
  • Market Maker (Aug 13) did a deep dive on CoreWeave as the poster child: a $104 billion backlog, up 246% year over year (now ~$130B after another $25B of commitments), against $9.4 billion of capex in a single quarter on ~$2.5B of revenue, and for scale, Apple's quarterly capex is ~$2.5B. Their verdict: "it's a funding risk above everything else." The bull answer to the "chips are obsolete in a year" worry is NVIDIA's CUDA "value cascade", where older chips move down a hierarchy of workloads and still earn revenue for 5–6 years. Michael Burry, of Big Short fame, closed his fund partly betting the other way (that chips only have a 12-month useful life).
  • The T. Rowe manager, notably, is not worried about capex slowing, he expects hyperscaler capex to accelerate to ~$1.5–1.6 trillion in 2027, pointing to Amazon's disclosure that short-life AI assets "break even in roughly two to three years and then... last five to six years," AWS growth reaccelerating to 37%, Azure to 43%, and Google Cloud growing 82%, all at ~50% incremental margins.

2. Key Debates

Debate A: Is SaaS Actually Dead, or Just Changing?

Not dead (TBPN, Freshworks) vs. structurally broken pricing (the sizeocalypse camp). The wedge: everyone agrees the cost to build software has cratered and that monolithic code is a weaker moat than it was a decade ago. The optimists say durable strengths (data, distribution, network effects, low share-of-wallet) protect the winners, and the sell-off threw babies out with the bathwater. The pessimists say the billing model itself (seats and predictable recurring revenue) is what breaks, and that undermines the whole valuation framework. Freshworks' synthesis: wrong prediction, right patient.

Debate B: Application Software vs. Infrastructure Software

App software in trouble (T. Rowe) vs. still defensible. The bear case is that the chat/agent interface strips value away from the app layer and hands it to the frontier labs; the more constructive view (see Palantir below) is that vendors who own proprietary data, deep workflow integration, and outcome-based contracts can thrive. Infra/consumption names (Snowflake, Datadog) and data-gravity names (SAP) are seen as more insulated.

Debate C: Anthropic's ~$2T IPO, Visionary or Bubble?

This was the debate of the week.

  • Bull: On All-In (Aug 14), Gavin Baker called Anthropic "an exceptional business" and predicted that even if it prices at $2T, "it might trade" to $3–4T; he'll "take the over" on Wall Street's 2027 revenue consensus by 25–30%. David Sacks agreed ~$400–500 billion of exit ARR next year "seems very comfortable" (from ~$100–120B exiting 2026), while flagging that the real limit becomes physical: "is there enough compute? Is there enough energy?" Both were skeptical of the internal hubris, with Sacks recounting that Dario Amodei has reportedly said Anthropic "might be the only private company in the world at some point."
  • Bear: On The Exchange (Aug 13), Roger McNamee argued the labs are structurally unprofitable and moat-less: "there's not a single customer out there who pays more for an AI product than it costs the vendor to provide it," the industry has "invested $1.4 trillion in capital equipment... we are still in the first order of magnitude of cost reduction," and he'd tell Meta to "shut down its AI operation."
  • Skeptical middle: On Motley Fool Hidden Gems (Aug 14), Jason Moser called OpenAI and Anthropic "one-trick ponies in a market that seems to become just more and more commoditized by the day," while Lou Whiteman conceded "I know it can't last forever, but gosh, it works right now."

Debate D: The Whole Market Is a Bet on Two Companies

  • Eisman's "Achilles heel" framing on The Real Eisman Playbook: research suggests ~70% of hyperscaler AI revenue and 25–35% of total cloud revenue comes from OpenAI and Anthropic, and "Oracle has a $600 billion backlog, and half that backlog is from open AI alone... both companies lose billions and are reliant... on raising capital for their survival." If a price war breaks out (cheap Chinese open-weight models are the trigger), "the hyperscalers may have to pull back. And if they have to pull back, the entire AI chain goes into reverse."
  • On Prof G Markets (Aug 10), Scott Galloway sharpened the math: to justify current capex, AI needs to "create $2.5 trillion in incremental revenue... Right now, the cumulative revenue being produced by AI is $150 billion. So we need it to 15x." He also flagged that Chinese LLMs went from ~10% of token usage in January 2025 to "approaching two-thirds," and that Anthropic just borrowed "$36 billion from Blackstone this week." His stance: AI will be transformative long-term, but "there's more asymmetry to the downside than to the upside" over the next 24 months.

Debate E: OpenAI vs. Anthropic, Who's Winning Right Now?

  • On AI Chat (Aug 11), host Jaeden Schafer flagged that OpenAI's $7 billion employee share buyback came at an $852 billion valuation, the same mark it held back in March, reading it as a sign of stagnation and "they're not in a big rush to go public," while "Anthropic just absolutely mopped the floor with their growth." But he expects OpenAI to fight back, having personally switched from Claude Cowork to ChatGPT Work.
  • Sacks (on All-In) said OpenAI has pivoted hard into coding and is "now... growing as fast as Anthropic," with month-over-month growth "now over 20%." A telling data point from Tech Brew Ride Home (Aug 13): Anthropic's flagship Fable 5 captured only ~6% of tokens in its first month because it costs roughly twice as much as OpenAI's GPT 5.6 Sol, a possible "new ceiling on what companies are willing to spend on AI."

Debate F: Who Captures the Value in the Agent Economy?

The Excess Returns manager sketched an "80-20 world": ~80% of AI tokens will be open-source or cheaper "good enough" models, but frontier labs capture "65 to 85% of the economic value" by orchestrating them, comparing coding's TAM ("$300 to $400 billion") to "all of application software." The open question he flagged: do OpenAI/Anthropic capture that, or does a Microsoft/Palantir layer sitting on top and model-swapping win instead?

3. Specific Names, Bull and Bear Stances

Palantir (PLTR): three very different views.

  • Strong bull: Gil Luria (Head of Technology Research, D.A. Davidson) on Pitch The PM (Aug 11) upgraded the stock when the valuation compressed from "200 times cash flow" to ~50x forward cash flow while growth accelerated to 93% (from 85%): U.S. commercial +150%, government +90%, 62% operating margins, 157% net revenue retention from outcome-based pricing. His moat argument: the "ontology" (mapping messy company data to how a specific business actually works) plus forward-deployed engineers is extremely hard to copy: "C3.ai... tried to replicate the model. And it's an absolute disaster." He thinks it's "pretty straightforward to underwrite 50% to 70% CAGR for at least three years, probably five years."
  • Bull, but take profits: On The Morning Filter (Aug 10), Morningstar's Dave Van Knapp called the quarter "just phenomenal" (record 62% operating and 63% free-cash-flow margins), but with the stock up ~50% since his pick and now at a ~12% premium to a $153 fair value, "I wouldn't argue against taking some profits here."
  • Bear: Via Saxo Market Call (Aug 13), Michael Burry is reportedly short, buying March 2027 puts and (in a figure the host found "a bit excessive") suggesting shares are "worth under a dollar," citing 31 million shares of stock-based compensation, off-balance-sheet commitments, and infrastructure purchase obligations "set to triple."

Anthropic (private): the week's center of gravity. Reportedly targeting a ~$2 trillion IPO in October (last valued at $965B; ~$47B annualized revenue run rate as of May; investors expect $100–120B annualized by end of 2026, ~800% growth per Tech Brew Ride Home). Signed a 20-year, $9.1 billion / 191-megawatt compute deal with former Bitcoin miner Riot Platforms (The Rundown), and is in talks to buy Israeli chip-optimization startup Decart for ~$6 billion, its largest acquisition to date (Bloomberg Intelligence, Aug 13). Bull: Gavin Baker (All-In). Bear: McNamee (The Exchange) and Eisman. Investors are pressing three risks: cheaper Chinese models, tension with the Trump administration, and data-center backlash (The Rundown).

OpenAI (private). $7B employee buyback at an unchanged $852B; reaccelerating in coding (GPT 5.6 Sol at ~25% of tokens); not planning to IPO until 2027 (AI Chat).

Databricks (private). Closed a $5B round at a $190B valuation, crossed a $7B revenue run rate, +80% YoY; its Lakebase database for AI agents already passed $100M run rate and its Lakehouse tool $1.5B (Tech Brew Ride Home). CEO Ali Ghodsi: "demand is crazy."

Canva (private). The cautionary tale of the week on 20VC (Aug 13). Growth decelerated from 30% to ~20% at ~$4B ARR, partly because frontier-model API costs made it "too expensive to effectively be subsidizing users." Jason Calacanis pegged it at "probably worth $12 billion right now," with the real question being existential: "is AI going to be a feature they can incorporate? Or is it a new thing that makes them obsolete?" The panel's fear: the prosumer segment is the most exposed because "everyone's used ChatGPT... our agents never even suggest these products." They contrasted it with the "no existential question" infra names (Datadog, Cloudflare) trading at 15–17x forward revenue.

Adobe (ADBE). On The Investor's Podcast (TIP837, Aug 13): the stock is down ~70% while revenue still grew ~10% a year, as the multiple compressed from "22 times earnings to 11 times earnings." Bull case: enterprise customers want full creative control that AI's probabilistic output can't guarantee. Bear case (why the hosts sold): Adobe could lose the long tail of non-studio users and the top-of-funnel of young professionals to cheaper AI tools, and "you got to ask yourself whether you want to bet against AI right now in the first place."

Workday (WDAY). Reportedly in take-private talks with Silver Lake; stock jumped 15–20%. On The Josh Bersin Company (Aug 14), Bersin is bullish: the SANA and Paradox acquisitions make its "AI strategy... exceptionally strong," the HR software market is "very, very healthy," and its price-to-sales ratio is "half or a third that of the AI companies", and a PE firm could "double their investment in three to five... years."

Salesforce (CRM). The first visible big-enterprise pullback surfaced on The Information's TITV (Aug 12): the USDA is shifting ~$9M/year to C3.ai for correspondence tracking, and its big Salesforce agreement is set at $277M for 2025–2030, down from ~$294M the prior five years. Reporter Laura Bratton: "investors are scared that... this could become a broad trend and threaten Salesforce's pricing power, which is why the stock is down so far this year." Separately, the Elon Musk Podcast framed Salesforce's ~$3.6B purchase of Intercom's Finn agent as a possible "panic acquisition." Product-side bull: Salesforce's Matt Kravitz walked through Service Cloud's three levels of AI-agent maturity on The AI in Business Podcast (Aug 12).

ServiceNow (NOW). A Morning Filter (Aug 10) buy: Morningstar's 4-star, ~24% discount to fair value; deals with 5+ AI products grew 5.5x year over year, over $1B in annual contract value, and analyst Dan Romanoff calls it "one of the best blends of growth and margins in enterprise software... AI [is] not hurting the firm's fundamentals."

Shopify (SHOP). Q2 revenue $3.58B (+34%), GMV $116B (+32%), operating income +68% on flat headcount with more than half of May's code written by AI. On The Watson Weekly (Aug 10), Rick Watson is bullish on the operating leverage (AI orders tripled year over year, Sidekick daily active merchants up 3.6x, ~34M merchant conversations) but bearish on disclosure: "Shopify still has not given anyone a dollar figure for what Agentic Commerce contributes to its GMV."

Cloudflare (NET). On Motley Fool Hidden Gems (Aug 11), Jon Quast highlighted CEO Matthew Prince's claim that AI agent traffic surpassed human activity on Cloudflare's network in May and could be "1000 times bigger than human traffic within the next five years." Matt Frankel and Tyler Crowe were skeptical of the 1000x figure (agent traffic grew ~18x in the past year), but agreed the direction is real and could upend the ad-funded internet business model.

CrowdStrike (CRWD) & Palo Alto Networks (PANW). Repeatedly cited as clear AI winners in security (TBPN): PANW +121% ($320B), CRWD +107% ($230B); PANW CEO Nikesh Arora's $10M dip-buy is now worth ~$26M.

Snowflake (SNOW) & Datadog (DDOG). Held up as the "no existential question" infrastructure comps: mid-20s growth, 20%+ operating margins, ~15–17x forward revenue (20VC), and, per the T. Rowe manager, the safer place to hide within software.

C3.ai (AI). Won the USDA task from Salesforce, but revenue "fell by more than half" last quarter and returning CEO Tom Siebel called the results "unacceptable"; hosts speculated it becomes an acquisition target (The Information's TITV).

CoreWeave (CRWV). The clearest read on the neocloud debt machine: $104B backlog (+246%), $9.4B quarterly capex, ~$567M adjusted net loss, scaling power from 1.5 to 4.2 gigawatts (Market Maker); Eisman noted revenue +112% and the stock +19% on the print.

Also flagged: IBM shed "$69 billion in market value in a single day following a profit warning," and SAP's Hybris e-commerce platform hit end-of-life (Elon Musk Podcast). On vertical SaaS, Leaders In Payments (Aug 10) noted 75–80% of vertical software companies have already embedded payments, with AI-built platforms entering verticals fast and creating winners and losers based on partner choice.