TechStockFundamentals

I think there are basically 2 prevailing strategies in the market. 1) Earnings revisions Only two questions matter near-term: Is the story getting better or worse, and are numbers going up or down? That's it. For momentum names, in either direction, that's all that matters -> is the story continuing to improve w/ beat and raises, or is the story continuing to erode with misses / guidance reductions. There is more complexity at extremes (ie second derivatives) that also works -> the story has stopped getting worse and numbers are no longer going down so upside optionality, or narrative no longer improving and downside risk to numbers so downside optionality. Earnings revisions work because they are reflexive. The company is doing well, estimates are moving higher and the stock is rising. More investors do work on the name and the due diligence will be positively biased as they generally ask more favorable questions to understand the strengths of the company. This leads to more buying and the increased stock price reinforces the bullish narrative. Hopefully, management is smart enough to set forward guidance at beatable levels and the step-ups continue. Same is true for reflexivity on the downside. A company is struggling and missing its own guidance and expectations. The stock moves lower as obviously more people sell than buy. Investors do more work but are naturally more focused on the reasons for corporate underperformance and corresponding risks. Further stock price declines fuel the negative narrative which feeds on itself. Companies often don't take their medicine as far as fully disclosing bad news and the negative revisions continue. So again, the playbook for this is simply 1) is the story getting better or worse?, and 2) are numbers going up or down? 2) Valuation From time to time, stocks get too cheap or expensive relative to fundamentals such that the disconnect is very interesting (let's say can make at least 50% to get back to conservative valuation target in either direction within 12-18 months). Very often, large disconnects occur when earnings revisions are unclear to everyone (stock is really cheap but no one has any clue when numbers stabilize, stock is really expensive but everyone thinks company will continue beating / raising). However, the insight here is that valuation is interesting in and of itself for the right kinds of companies (solid business, transitory issues, acquisition candidates, etc.). And that no one has visibility into near-term numbers is already in the stock and why the valuation is so compelling. This opportunity also always arises in tough macro periods where everyone is frozen bc who can predict how the great financial crisis or covid pandemic plays out - so that uncertainty is shared by everyone and more than represented in stock valuations. And that is why the opportunity exists - because everyone is worried about the same imponderable thing and it's already in the stock price. The only advantage you need here is a deep understanding of and conviction in the business, and freedom / willingness to buy it for the long-term (and the returns never take as along as you think). Reality My experience is most investors typically only pursue one of above strategies (and might only be allowed to pursue one or the other). So they won't / can't buy $WDAY at 20x FCF bc they think guidance might be lowered (company misexecution, tough macro, etc.), or they won't sell $WDAY at 35x FCF bc they are confident everything is great and company will continue beating. I also have found earnings revisions investors tend to have nearer-term horizons and be less valuation sensitive. This is because it's hard to fill a portfolio with names that will both beat earnings and trade at reasonable valuations. So, on balance, these investors will compromise on valuation. Valuation investors have longer-term horizons and are by definition valuation sensitive and less sensitive to revisions. This is because it's hard to fill a portfolio with names that are both cheap and that will beat earnings. So, on balance, these investors will compromise on earnings visibility. Optimal I have found earnings revisions offer the most explanatory power near- to mid-term. I want to be long stocks where numbers are moving higher and short stocks where revisions will be lower. But I also want think it is an advantage to be able to cast that strategy aside and buy companies at valuation extremes. This is when a good company is "too cheap" even if I'm not sure of near-term revisions. There are other criteria that need to be in place for this to work -> solid business w/ reasonable estimation capabilities, potential negative revisions are manageable and likely in the stock, less risk of permanent impairment over a 12-18 month horizon, balance sheet in order and not indebted, etc. Finally, the intersection of the two is where you can nail inflection points in businesses / stocks. A company is too cheap and is about to start beating numbers which will drive both estimates and the multiple higher. That's where the big money resides. Anyway, this overall thinking is why I structure my notes as follows: Is this a bad, good, or great business? Is the story getting better or worse? Are numbers going up or down? What is the risk / reward?

Jan 26 fear was just semi guys saying sw is dead. And sw stocks all ended up rallying pretty hard. This fear is semi and model guys saying sw is dead. And having new tools to show for it. I mean there is no AI product out there to replace CRM, WDAY or NOW wholesale and today - even if you wanted to. So people are scared of something that doesn’t exist yet and that customers don’t want to do.

@TechFundies Couldn’t you literally have said this at end of Jan 2026 or do you think “this time is different” and if so, what changed? genuinely curious

The interesting thing about SaaS is the market basically went through a gut-wrenching process where it put every possible AI fear on the table culminating with "we're all out of jobs". The downward price movement emotionally reinforced the validity of these concerns into investors' minds. And when stocks are going down quickly, it allows anyone to say whatever bearish thing comes to mind without getting laughed out of the room (and vice versa when stocks are ripping). But now we are perhaps at a point where any investor who holds shares in SaaS has digested these concerns and already determined they are in. In other words, I would think most people who own SaaS at this point are much firmer hands than a few months ago. Like what needs to happen from here to get current investors to sell? And is that going to happen shortly? Will be interesting to see how it plays out from here.

Comments on this post are very interesting. Lots of people chiming in about how they used Claude Code to build apps that have never been available with their domain expertise. Seems to me that might be the most powerful outcome. More customized sw for the long tail.

Head of Product Marketing at $MNTN and previously was Ad Growth Marketing at $ROKU leaving 6/22. Pretty constructive on the business - differentiated technology across targeting / attribution, and right place / right time as far as CTV ramp. Highlights -Head up product marketing for Quickframe – responsible for customer segmentation, perosna build out, positioning, and messaging. -Majority of new customers are new to TV and spend pretty lavishly on search / social so care about lower funnel outcomes. Lots of demand – so many smaller businesses were told they needed to have six-figure spend for ad creative and six-figure spend for campaign and were locked out of traditional TV. Early days. -We position CTV as performance play w/ healthy brand side effects. -Inventory – Have direct relationships w/ >150 premium publishers. -Targeting – customers can use first-party data like retarget website visitors or email addresses. -Have patent-pending attribution system called Verified Visits that lets advertisers know a targeted hh has been served and whether conversion in a window happens. -Have partnerships w/ Northbeam / Haus / Rockerbox that help clients figure out attribution in a window between MNTN, GOOGL, etc. Attribution can be described as a bit of religion – can come down to what you believe. -MNTN attracts a lot of mid-market advertisers as well as SMBs that spend hundreds to low thousands per campaign. Also have hundreds of agencies that use MNTN for their CTV practice. -Have systems and teams in the back-end to nurture install base. There is an educational journey that goes along w/ the advertiser’s performance journey. MNTN offers fairly substantial knowledge base for self-help but also people available to help advertisers. Competition -TTD and AMZN don’t really come up. TTD built for biggest agencies. -tvScientific / PINS has similar pitch. MNTN tech is more advanced. Some value in having PINS data and retargeting on TV. -Vibe – much newer entrant and more focused on very long tail of SMB. Customers say they don’t have audience-building or measurement precisions they expect.

@undrvalue 100%. I’ve been doing enterprise and channel checks for >20 years so the difference in buying behavior is visceral for me. But totally appreciate it is not for most investors, and that perhaps things will change on margin going forward.

@TechFundies I think a big part of the disconnect in narrative and fundamentals is few people understand enterprise behavior versus startup/individual behavior

Interesting $CRM / $NOW / $WDAY check with large company in the alcohol space. Another large company that prefers to buy AI via SaaS incumbents, and highlighted their failed Open.AI efforts. I have to include the $CRM coversheet updated for the accelerated buyback bc it's so interesting. Not even giving that much credit for AI so just sustaining 9% organic cc yy subscription growth w/ some margin expansion. Results in ~20% GAAP operating profit growth which is 50% faster than EPS growth in QQQ. Trades at 16x profits (includes SBC and removes amortization, gives credit for strategic investments). So now stock potentially benefits from leverage in 1) operating margins, 2) multiple, and 3) financial leverage. Going to be one hell of a case study in either direction. Highlights -AI strategy is give preference to built-in agents in big SaaS vendors – WDAY, NOW, CRM, etc. CRM -Use consumer goods cloud, data cloud, marketing cloud (promotional assets, etc.) -Use data cloud with data from retailers, distributors, etc. to look at sell-in and sell-out data -Using AF to query data – where is their inventory gap, etc. Resolving this quickly is big ROI as far as driving revenue. -Using AF to take pics sales reps take of shelves and then do audit – is product missing at eye-level when running a promotion, etc. AF will automatically let distributor know they didn’t fulfill promise when running a promotion, big banner should be outside, etc. Reduces week of work to instant. -Started these 2 POCs last year, lasted 4 months and now going into production piece by piece -ROI definitely there – if can save 10 hours a week per field rep across 200, then saving a big chunk. -Will pay 20% more on base license for AI. Consumption units could go up another 10-20% on top. NOW -Using for ITSM. AI improved first-call resolution for some tickets by 50% and reduced outsourced helpdesk by 10% (moved number of reps to tier 2 support). -ROI positive out of the gate and increased morale for users. WDAY -Use for talent recruiting, perf mgmt., comp, payroll, benefits -Using AI within HiredScore and Peakon for employee surveys. Lot of data / surveys / insights are AI impacted. -Employees in HR in some cases saving 30-40% of time from faster data collection from third parties on recruiting, open positions, etc. -ROI clearly there as far as employees moving their time to more productive tasks Open.AI -Worked on some projects but “gone out of control in terms of cost overruns”. -“What we built and how we built it and everything .. it’s not all there. It’s our fault but isn’t optimal.” -Had some regular ChatGPT premium enterprise licenses. This has fallen through and now moving towards [$MSFT] Copilot for enterprise search and building AI agents. Moving bc of relationship, price discount, etc. Multiple reasons – it’s just pre-built. Even free co-pilot works well for 80% of the population.

@matt_slotnick Sir. My account is named tech stock Fundamentals for a reason.

this is definitely a software narrative violation

Thank you for this and your contributions overall. Totally random thoughts: feel free to ignore entirely. -Accuracy: Is the analysis based on detailed financial disclosure from the companies or more so based on publicly disclosed / reported / rumored figures? Just helps calibrate standard error around various views like “From my outside in work, contribution margins are awesome.” -Similarly, the article also mentions “the data supports this” multiple times but it's unclear if "the data" is hard or soft. Again, just helpful to know as far as what conviction level to place around the comments. -“ Without getting into private data, I can say the dynamics of the labs training payback period doesn’t look that different from infra companies CAC payback periods.” Quick calcs on $MDB / $SNOW indicate $1 of average trailing 24 month S&M (excluding SBC) leads to $0.45-.50 of incremental cash product gross profit. So ~2 year paybacks at best since this excludes SBC and economic reality is more like 3 years including it. Understanding old models don't go away overnight, that seems kind of long? -“Said another way, if it takes you 6 months to pay back Model A’s training cost on a $1B revenue base, and Model B costs 4x more to train but you’re on a $3B revenue base… the payback period actually shrinks. The ratio is getting better, not worse.” I assume this 6 month payback is hypothetical since it is too distant from the 2-3 year calc in the prior bullet? -Taking the above bullet at face value: Let’s assume 50% GM. So Model A generates 1b in rev, 500m in GP / year and costs 250m to build (1/2 a year of training cost paid back) – so Y1 looks like 250m to train and 250m of profit (500m of gross profit for the year – 250m training cost). Model B costs 4x to train or 1b, generates 3b in rev, 1.5b in gross profit, and 500m of profit. So revenue tripled while profit doubled? Wouldn't that suggest payback period is extending (Model A paid back 100% in Y1 and Model B paid back 50%?) Maybe the explanation is gross margin is meaningfully higher for Model B because of scale? Or maybe this is just a very hypothetical example? Ok, like I said, feel free to ignore some or all of this:)

@tanayj I'm just surprised. I used the portal a decade ago and it was cleaner / had no errors.

@TechFundies PMs working on ads there would always joke how crazy it was that 100B of ads (at the time) were bought through ads manager given how bad it was haha

Good god. Has anyone ever placed ads with $META? I have a side project where I'm trying to help this guy get on his feet and start a business. Told him for fun I'd handle the digital advertising so I can leverage all the new AI stuff and see how $METAs ad machinery looks these days. Just trying to place a simple, targeted ad on $META and what a freaking headache: -IG / FB accounts should be linked but somehow that process is complicated and buggy -Even trying to just use the simplest options is pretty complex. Had to keep pasting screenshots into ChatGTP for guidance. -Lots of regular pages are very buggy. Buttons that don't work, submission forms that just spin endlessly, etc. -AI didn't seem to be embedded in the creative process at all which I found surprising. Even the auto-create video function was complicated to use / slow / buggy. I think I got it done but GOOD GOD that was not a pleasant experience.

I get all that. I'm just saying I was shocked to navigate their portal and hit multiple buttons that didn't do anything and get several errors that basically just said error without any remediation advice. Was shockingly frustrating for something that ought to be perfectly seamless. Sort of showed incompetence to me - not opportunity.

@HedgeyeComm @TechFundies They control the networks, you basically have to go through them, there is no opportunity

@LOGOinvestor Good stuff. Thx for sharing. Tell $NOW to send you your W-2 income for being a strategic sales rep:)

$NOW they cant seem to spell correctly!

@TechFundies Had fun dropping in a bunch of content from you and others into NotebookLM, interesting slides and synthesis! Once u get AI, ur never going back, I dont think people quite understand the implications of this, stocks certainly dont = the opportunity $NOW

@ZezimaLincoln I think knowing how all this plays out is an unanswerable question. What is known is that SaaS has incumbency advantage and AI offers the opportunity for the only meaningful product cycle the industry has seen in a very long time. Might be rising tide lifts all boats.

@TechFundies This is consistent across all the big SAAS companies. But why will NOW orchestrate or govern agents? Everyone is chasing this aspiration and MSFT is probably better positioned to do so

Contact is director of strategic accounts at large NA service integrator focused on fin svcs and insurance sub-segment. Constructive on $NOW. 0 evidence put forth to support SaaS bearishness. Kind of the opposite actually - expects to accelerate $NOW business in 2026. Highlights -NOW considered to be mission-critical platform by most customers. For most incremental IT / business needs, customers first ask whether NOW should be used before considering alternate vendors. -NOW is still developing their insurance-specific solutions – will get there but a bit of a feature gap for now as they work towards maturity. -Top 3 solutions currently are security / risk which is in very high demand, IT tech workflows and CSM. -Are seeing shadow-IT where business units doing AI proof-of-concepts and pilots bypassing IT. This just causes a bit more IT delays as they get their arms around this spend, whether to bring those projects to IT to solve, etc. Think this will get resolved and accelerate spend over the year. -Lot of customers have AI sprawl and need way to orchestrate and govern these silos of AI automation. Looking to NOW to solve this. -See NOW sometimes beating CRM for AI governance in customer service mgmt. Won bc got IT to view the deployment as enterprise-wide governance of AI, and not CRM specific. -Moveworks is a very compelling stand-alone platform and a compelling acquisition. Basically is better agentic in-case solution for a lot of employee experience use cases. -Today see customers asking for AI solutions from NOW as much as NOW is pushing it – market just maturing -Customers always push back on price (20-25% increase for higher tier) but solvable. Just need more conversation up front and alignment around ROI. Once they understand that, not pushing back as much. -NOW leaning more heavily on partners in 2026. 1) Need help selling / deploying AI apps along w/ maintenance, 2) NOW has sold some shelfware and need partner help to get those modules deployed so clients get value from them and not churn risk. -Business grew 25% in 2025. Think will grow 25-30% this year as NOW benefits from companies leveraging them to get AI sprawl under control.

@jake_ayes Large HVAC equipment supplier

@TechFundies Can I ask what sector?

Contact works at large enterprise leading AI deployment. Punchlines are 1) They are focused on adopting AI through their existing SaaS vendors first ($MSFT, $CRM, $NOW, $WDAY), and 2) the SaaS companies are finally establishing AI product-market fit. If you can mentally separate a company from its stock, it's really quite wild how constructive fundamentals are becoming. Highlight -More aggressive in deploying all market solutions for AI – want to launch all AI features from existing platforms. -$CRM AF – company claims they have 150 pre-built agents. Have enabled 10 – B2B sales pipeline mgmt. (RFP responses), CPQ agent that puts together proposals, claims / deduction mgmt. agents, translate manual fax orders (get close to 1k orders each year through faxed manual orders which now use AI OCR to input). -Changed pricing to action which is more palatable. Using pay-as-you-go w/ pre-pay discount upfront. -Big advantage in being completely native to CRM – already have entire sales teams running on it. -Biggest task is to get all the data integrated properly. Mostly doing tasks that involve CRM data bc piping in other app data is technically difficult. -Don’t use data cloud and just use local data / AG needs; for CPQ grabbing some data from SNOW -Unlocking a lot of AI value from sales functions. Automated CPQ allowing reps to bid on far more projects and gaining share from competition. Actually seeing decline in win/loss bc bidding on more projects but is really an increase in net wins. -Positive ROI bc do AB tests on reps when rolling out and can see it -Challenge w/ over the top agent is providers don’t allow for full access via MCP. So CRM might have 50 data elements but only make 30 available. So sticking w/ app-based AI. -Have unused seats on CRM and company letting us move those dollars to AI consumption. Near-term spending more on AI bc can’t get out of excess wasted seat spend. On renewal will see where they land as far as potentially reducing seats and spending more on AI. AF will be 8-10% of total contract. -MSFT Copilot – 25k licenses and use studio. Made API calls entirely free in Copilot Studio which is big deal. -NOW - Buying Pro Plus tier with Assist packs. Use ITSM, ITOM, HR and one more -Customer service is doing 60% deflection across ITSM. -WDAY – was a laggard for a while but HiredScore and Paradox are good products and being adopted.

@HarshilOs At this point, why bother if Excel AI just works?

@TechFundies You should be using Claude plugin in excel rather than using cowork.

I tried using $MSFT Excel in Agent mode to update my $MDB model using a press release and a detailed prompt. And it worked! It's a simple model but it pulled every figure accurately including operating metrics, figured out where I was collapsing numbers (cash = st cash + investments), pulled forward formulas, highlighted cells it couldn't fill because data was unavailable (usually disclosed in transcript or Q). I'm not sure if it saved me a ton of time but did save me from having to do it which is certainly nice. I just double-checked it which went rather quickly. Hmmm, pretty interesting. Random thoughts -Model selection menu suggests this was powered by Open.AI but whatever. I don't care what engine is needed, I just want my car to work (Excel, Salesforce, etc.). $MSFT can figure out which model to use and it is better for me to have them do it so I don't build separate workflows on a potentially losing model provider. -I still need to work in Excel. Yes, AI helped update the model but I now need to look at the numbers and adjust my estimates in a tool in which I am highly productive. AI doesn't at all make me want to use Excel less. In fact, I might use it even more but for more productive use cases (adjusting forecasts takes share from inputting numbers). -I would definitely pay for this. I'm on a business basic plan but for some reason have access to premium features. Whatever, the point is I'm very happy to pay more than the $7.84 / mo I currently pay to $MSFT for this. It's the first "no-brainer" purchase decision MSFT can put in front of me in a long time. -I'm not running a fund currently but, if I were, the compliance / security functionality available in the $MSFT admin portal would be important and required. Can't fool around with SEC / 1940 act regulations. You ever been through an SEC audit? I have multiple times - no fun. Now, on the other hand, SaaS is dead so maybe I'm going about this all wrong:) (By the way, I ran the same exact prompt in Claude Cowork on a folder. It ran out of credits on my $20 / mo. plan)

I tried Claude for Excel a few weeks back and it didn't work at all. But I gave it a simpler prompt and that was also a few weeks back so might be better now. But, as a point on user preference, I'm not really going to go back and test it repeatedly now that $MSFT seems to work. Why would I? That's my point on SaaS incumbency and AI functionality.

@TechFundies @SouthernValue95 Also, how did you give it the data to update the excel model?

@HanchungLee $MSFT worked and it did both GAAP and non-GAAP updates correctly.

@TechFundies w8, lhs is excel, rhs is claude cowork. which one worked? and is it a presser? and how well it consolidated gaap nongaap and the typical metrics resolutions like fcf or ebitda?