Freshworks at $8: Google Holds, Insiders Buy, Nobody Says Sell

The Largest Position in Alphabet's Book

As of March 31, 2025, the largest stock position disclosed in Alphabet's 13F wasn't an AI lab, a chip designer, or a satellite play. It was Freshworks - a customer-service software company founded in Chennai - at $228.7 million, 14.5% of the entire reported portfolio.

Today Freshworks closed at $8.04, down 57% from its January 2025 peak. Alphabet hasn't sold a share. The CFO put $2.0 million of his own money into the stock in November. The CEO added another $994,000 five weeks ago, three dollars lower. And among the twenty-odd analysts covering the name, not a single one rates it a sell - although one of them just told clients it's worth almost exactly what it trades for.

We used Barebone to pull every Alphabet 13F since 2021, six months of Freshworks insider filings, and every analyst action since December. What follows is a study in three signals - the owner, the insiders, the street - on a stock the market has left for dead.

16,206,643 shares Alphabet won't sell. $3.0 million of insider buys against $558,000 of sells. Eight price targets, zero sells, a median 68% above the tape. The receipts, in order.

Anatomy of a Left-for-Dead Stock

First, the crime scene. Application software is in a rolling crash. The trigger was agentic AI - autonomous tools, led by Anthropic's push into "AI coworker" territory, that investors fear will do the work enterprise software charges per seat for. By February 6, CNBC was asking whether the selloff was an "illogical" panic or a "SaaS apocalypse."

The damage is not subtle. The IGV software ETF is down 24.6% year-to-date ($105.69 to $79.75) and closed today within a dime of its January 27 panic low. Ten weeks in, the sector still hasn't bounced.

Freshworks fell harder, in two distinct legs.

Leg one was the long 2025 grind: from a weekly-close peak of $18.60 in late January 2025 down to the $11 - 13 range by autumn, as growth investors drifted away from mid-cap SaaS. Leg two was the air pocket: from $12.25 on December 31 to a weekly close of $7.24 by mid-February - a 41% drop in six weeks. The stock has flatlined around $8 since. Year-to-date it's down 34%; from the 2021 IPO price of $36, down 78%.

Freshworks makes customer-support and IT software - Freshdesk for service tickets, Freshservice for internal IT - sold mostly to mid-market companies. Named customers include Sony and Bridgestone. It is, in other words, exactly the kind of seat-priced application software the AI-agent thesis says gets eaten first. Keep that in mind: it's both the reason the stock is at $8 and the caveat hanging over everything below.

Signal One: The Owner Who Won't Leave

Start with the most patient name on the shareholder register. CapitalG - Alphabet's independent growth fund - co-led Freshworks' $150 million Series H in November 2019, alongside Sequoia and Accel, at a $3.5 billion valuation. Freshworks IPO'd at $36 in September 2021 and has not seen that price since its first months of trading.

Here is Alphabet's disclosed Freshworks position, from the 13F filings:

Filing (period end)	Shares	Reported value	Implied price
Q4 2021	4,035,595	$106.0M	$26.26
Q1 2022	4,035,595	$72.3M	$17.92
Q1 2025	16,206,643	$228.7M	$14.11
Q3 2025	16,206,643	$190.8M	$11.77
Q4 2025 (latest)	16,206,643	$198.5M	$12.25

Two things jump out. Somewhere between 2022 and early 2025, the disclosed count quadrupled to 16,206,643 shares - most plausibly pre-IPO stock converting from unreportable Class B into reportable Class A, rather than open-market buying; 13Fs don't say. And then the count froze. Across the March, September, and December 2025 filings - through a 43% drawdown - Alphabet's position didn't move by a single share.

The version of this story you'll find on stock-data apps says Google bought $200 million of an $8 stock at a $22 average cost. The filings say something more precise and less cinematic. The $200 million is the position's value at year-end ($198.5 million), not what Alphabet paid. The "$22 average cost" is a vendor estimate seeded by where the stock traded when the shares first surfaced in 13Fs - $26.26 at the end of 2021 - not a number from any filing. What Alphabet actually paid traces back to a 2019 venture round and is unknowable from public documents.

What is knowable: Alphabet holds just under 6% of Freshworks, worth about $130 million at today's close - down nearly $100 million from the March 2025 mark - and the latest filing, signed February 12, 2026, shows no selling whatsoever.

Be honest about what that's worth as a signal. $130 million is a rounding error on Alphabet's balance sheet, and venture funds are structurally slow sellers. The signal isn't that Google is pounding the table; it's that the best-informed pre-IPO investor on the cap table has had four years of quarterly chances to quietly walk away - including two filings during the crash - and hasn't.

Signal Two: Executives Writing Personal Checks

13F holders inherit positions. Insiders choose them. Here is every open-market transaction by a Freshworks insider from October 1, 2025 through today's close, from the Form 4 filings:

Date	Insider	Role	Type	Shares	Price	Value
Nov 11, 2025	Tyler Sloat	CFO	Buy	171,615	$11.62	$1,994,166
Nov 17, 2025	Philippa Lawrence	Chief Accounting Officer	Sell	644	$12.05	$7,760
Dec 2, 2025	Philippa Lawrence	Chief Accounting Officer	Sell	5,846	$12.06	$70,503
Dec 5, 2025	Mika Yamamoto	Chief Customer & Marketing Officer	Sell	15,012	$13.00	$195,156
Jan 6, 2026	Philippa Lawrence	Chief Accounting Officer	Sell	765	$11.61	$8,882
Mar 2, 2026	Dennis Woodside	CEO & President	Buy	125,000	$7.95	$993,750
Mar 4, 2026	Mika Yamamoto	Chief Customer & Marketing Officer	Sell	32,577	$8.45	$275,276

It would be tidy to report that every insider trade in the window was a buy. Not quite: five sells cleared the tape, all from officers outside the CEO/CFO pair, several small enough to look like routine, programmatic selling around vesting. (Tax-withholding share surrenders, which aren't trades at all, are excluded.)

But the dollar math is lopsided. $3.0 million bought against $558,000 sold - more than five dollars of buying for every dollar of selling - and the two largest checks belong to the two people who see the numbers first.

The sequencing matters more than the totals. Sloat bought $2 million worth at $11.62 in November, before the AI panic - he's down 31% on that purchase. Woodside bought $994,000 at $7.95 on March 2, near the bottom of the crash, and is roughly flat. The CFO has been wrong for five months and hasn't reversed; the CEO looked at the post-crash price and added seven figures.

Insider buying is the only signal in this story where someone volunteered to be long at these prices with money they could have kept. It's also the easiest signal to over-read: a million dollars is conviction-sized for most people, and it is also excellent optics for a CEO whose stock is down by half. Treat it as a weight on the scale, not a verdict.

Signal Three: A Street With No Sell Button

The sell side spent February marking Freshworks down - without a single firm walking to the exit. Every rating action since December:

Date	Firm	Action	Target
Dec 17, 2025	BTIG	Initiated: Neutral	-
Jan 22, 2026	Oppenheimer	Outperform, maintained	$19 → $18
Feb 3, 2026	Piper Sandler	Overweight → Neutral	$20 → $12
Feb 11, 2026	Needham	Buy, reiterated	$15
Feb 11, 2026	Cantor Fitzgerald	Overweight, maintained	$15 → $12
Feb 11, 2026	Citizens JMP	Market Outperform, maintained	$27 → $16
Feb 11, 2026	Wells Fargo	Equal-Weight, maintained	$13 → $10
Feb 11, 2026	Piper Sandler	Neutral, maintained	$12 → $10
Feb 23, 2026	Jefferies	Buy → Hold	$20 → $8

Morgan Stanley sits at equal-weight with a $15 target. Tally the live ratings and the split runs roughly half buy-equivalents, half holds - and zero sells, on any tracker we checked.

The Jefferies move is the one to study. On February 23, Brent Thill's team refreshed its entire application-software coverage around a new AI-risk framework and downgraded four names; Freshworks took a cut from $20 to $8 - a 60% haircut to the target, landing within four cents of where the stock closed today. That's not an adjustment. That's a capitulation, dressed as a price target.

Which makes the overall distribution strange. The most bearish published target on the street is the current price - and the median of the eight live targets sits near $13.50, about 68% above the tape. When the floor of Wall Street's expectations equals the market price, one of two things is true: either the street hasn't capitulated enough, or the market has overshot everything the fundamentals justify.

A cynic would note that "hold with a target at the tape" is how the sell side says we have no idea without saying it. Fair. But notice what's absent: nobody - not even the firm that built an AI-risk framework specifically to find software's victims - is willing to argue this business is worth less than $8.

The Business Under the Stock

So what does $8.04 actually buy? The company reported fourth-quarter and full-year results on February 10:

Metric	Result
Q4 2025 revenue	$222.7M, +14% YoY (+13% constant currency)
FY 2025 revenue	$838.8M, +16%
Net dollar retention	108%, up from 103% a year earlier
FY adjusted free cash flow	$223.1M (26.6% margin)
Customers >$5K ARR	24,762, +10%
Freddy AI ARR	Crossed $25M in Q4
Cash & marketable securities	$843.7M
2025 buybacks	$386.3M
FY 2026 revenue guide	$952 - 960M, +13.5 - 14.5%

2025 was Freshworks' first GAAP-profitable year as a public company - though the milestone leaned on a large fourth-quarter swing, so the cleaner read on the business is the $223 million of adjusted free cash flow. The employee-experience side (Freshservice, the IT product) crossed $500 million in annual recurring revenue. Management's stated ambition on the February call: $1.3 billion of revenue by 2028, with AI products contributing $100 million of ARR - "not just a feature in our products, it's a standalone revenue line."

Now the arithmetic the price implies. At $8.04, Freshworks is a roughly $2.25 billion company carrying $844 million in cash and investments and no meaningful debt. Strip the cash out and the market values the operating business near $1.4 billion - about 6.4× trailing free cash flow and 1.5× this year's guided revenue, for a business growing 14% with expanding margins and a buyback running close to $400 million a year.

Six times free cash flow is not a software multiple. It's a multiple for a business in terminal decline - a mall REIT, a legacy media asset. That is the bet the market is making: that the cash flows you can see are the last good ones.

The Bear Case Is Not Stupid

Here's the strongest version of the other side, because at $8 it deserves the floor.

Freshworks sits in the AI kill zone. Customer support is the single most automatable workflow in enterprise software. If AI agents deflect most tickets, companies need fewer support seats, and seat-priced software deflates with them. Jefferies' framework flags exactly this: real AI and competitive risk in the core customer-experience business. Even Freshworks' own Freddy AI is partly a hedge against its own pricing model - every deflected ticket is a seat that may not renew.

The deceleration is real. Q4 revenue growth was 14%, down from 22% a year earlier; guidance implies roughly 14% again in 2026, with no reacceleration in sight. The 108% net dollar retention print was a genuine bright spot - but retention is a lagging indicator, and the AI pressure on seats is a forward-looking problem.

Each signal has a soft spot. Alphabet's stake is venture residue, not an active bet - and note that Alphabet hasn't added a share either; "isn't selling" is not "is buying." The insider buys are real but small next to executive pay packages. And a street with eight holds-and-buys and zero sells mostly proves analysts hate downgrading into a hole.

In our VIX work we drew a line between knee-jerk panics, which historically rewarded buyers, and systemic breaks, which punished them for years. The entire question here is which one the software selloff is. If agentic AI genuinely replaces seat-based support software over the next five years, this is systemic - and 6× free cash flow won't act as a floor, because the cash flow itself will erode. Cheap is only cheap if the denominator holds.

What This Means

The three signals in this story share one property: they tell you how the best-informed parties are positioned, not what's true. The owner of record with four years of exit windows hasn't used one. The two executives with the earliest look at the numbers wrote seven-figure personal checks, one of them near the lows. The street has slashed its targets and still can't bring itself to say sell. Positioning is unanimous in one direction; the price says the opposite.

What would settle it:

• Alphabet's Q1 2026 13F, due mid-May. The first filing covering the crash quarter. If 16,206,643 finally moves - either direction - that's the headline number changing for the first time in four years.
• The Q1 report in early May. Watch net dollar retention against the 108% print and Freddy AI ARR against the $25 million mark. Those two lines are the AI thesis showing up - or not - in actual numbers.
• The buyback cadence. $386 million repurchased in 2025 against a $2.25 billion market cap; at these prices, capital returns do heavy lifting.
• The sector tape. IGV closed today within a dime of its January panic low. Panic bottoms get retested; systemic declines slice through.

At six times free cash flow, the market has already priced "wrong" as the base case for everyone in this story - Alphabet, the insiders, the street. The next quarter of filings will tell us whether the people with the most information were early, or just wrong on a longer fuse.

---

Data: Barebone | Sources: Alphabet Inc. 13F filings (SEC EDGAR), Freshworks Q4 2025 results release (Feb 10, 2026), SEC Form 4 filings (Oct 2025 - Apr 2026), Jefferies software coverage note (Feb 23, 2026), Freshworks Series H press release (Nov 13, 2019) | Prices as of April 8, 2026 close