AI Chatbot Stock Analysis Series: Five AI Chatbots Predict Figma (FIG) Stock — A Week of Earnings, Volatility, and Diverging Forecasts

Introduction

At TheDayAfterAI News, we are on a mission to understand how artificial intelligence is reshaping the way we interpret financial markets. As part of our ongoing AI Chatbot Stock Analysis Series, we asked five of the world's most prominent AI chatbots to forecast the price trajectory of Figma, Inc. (NYSE: FIG) over the five trading sessions from 18 February to 25 February 2026.

The timing was anything but ordinary. Figma had just delivered a blockbuster Q4 2025 earnings report on the evening of 18 February — one of the strongest software prints of the year — triggering a dramatic after-hours surge. Yet the broader software sector remained gripped by the so-called “SaaSpocalypse”: a historic sell-off driven by fears that autonomous AI agents would erode demand for traditional software tools. The result was a uniquely charged backdrop in which company-specific euphoria collided head-on with sector-wide anxiety.

One notable absence: Google Gemini. Despite three attempts to run deep research queries within 30 minutes of market open, Gemini failed to complete its analysis in time. As a result, this comparison features five chatbots rather than the planned six. We note this openly as part of our commitment to transparency in AI benchmarking.

The five participating chatbots were:

• Grok — developed by xAI (Elon Musk)
• Claude — developed by Anthropic
• ChatGPT — developed by OpenAI
• Perplexity AI — developed by Perplexity
• Copilot — developed by Microsoft

Each chatbot was presented with the same base data and asked to generate a structured five-day stock price prediction including: predicted opening price (Thu 19 Feb), predicted closing price (Wed 25 Feb), expected trading range, probability of a net increase, and supporting rationale. What emerged was a fascinating display of both consensus and division — and a rare window into how different AI architectures approach the inherently uncertain art of market forecasting.

The Catalyst: Figma’s Q4 2025 Earnings

Any fair assessment of the five predictions must begin with the earnings report that set the stage. Figma’s Q4 2025 results, released after the close on 18 February 2026, were extraordinary by any measure within the current software sector environment.

Key Q4 2025 Metrics

The stock closed at $24.19 on 18 February — already up 4.7% on the day — before surging to approximately $27–28 in after-hours trading, implying a gap-up open of roughly 14–17% on the morning of 19 February. The stage was set for a highly volatile week.

Market Context Going Into the Week

Understanding the predictions requires understanding the broader environment in which all five chatbots were operating. Several macro and structural headwinds created a uniquely complex backdrop:

• The SaaSpocalypse: The iShares Tech-Software ETF (IGV) had fallen approximately 30% from late-2025 highs. On 3 February alone, the S&P 500 Software Index dropped 13% in a single session amid fears that AI agents from Anthropic and OpenAI would erode software seat counts.
• FIG’s Own Decline: The stock had fallen 83% from its IPO-day high of $142.92 (1 August 2025) to a 52-week low of $19.85 on 4 February 2026 — a catastrophic drawdown that placed every major moving average well above spot price.
• Monthly Options Expiration (OPEX): Friday 20 February was monthly options expiration day, historically one of the most volatility-amplifying sessions of the month.
• Delayed Government Data: Both Q4 GDP advance estimates and December PCE inflation data — delayed by a late-2025 government shutdown — were scheduled for simultaneous release on OPEX Friday, creating an unusual “triple threat” of volatility catalysts.
• Hawkish Fed Posture: FOMC minutes released 18 February indicated several officials were open to rate hikes if inflation remained sticky. The federal funds rate sat at 3.50–3.75%.
• Geopolitical Tension: US-Iran nuclear talks were stalling, pushing oil prices to approximately $63–85 per barrel across different chatbot assessments.
• Nvidia Earnings: Scheduled for after the close on Wednesday 25 February — potentially the largest single macro catalyst for tech sentiment during the period.

The Five Predictions: At a Glance

The following table summarises the core numerical forecasts from each chatbot. All five were asked to predict the opening price on Thursday 19 February, the closing price on Wednesday 25 February, the expected five-day trading range, and the probability of a net price increase over the period.

Individual Chatbot Analysis

1. Grok (xAI) — The Optimist

Opening Prediction: $26.50–$27.00 | Closing Prediction: $29.00–$31.00 | Bullish Probability: 75%

Grok delivered the most bullish forecast of the group and also the most granular, producing a deeply researched report spanning technical indicators, options flow data, macro calendar analysis, fundamental catalysts, and international market conditions. Its 75% probability of a net price increase over the period placed it as the clear outlier on the optimistic end of the spectrum.

Grok’s case rested on several interlocking pillars: the post-IPO descent having created “extreme oversold conditions” with RSI recovering from below 30, a bullish MACD crossover forming, an unusually call-heavy options flow (put/call ratio 0.57), heavy dark pool accumulation signals, and a surging AI user base with net dollar retention at a ten-quarter high of 136%. It also cited $1.2M in premium placed on $30 calls expiring 13 March — an institutional bet on further gains.

Where Grok stands out is in the sheer breadth of its supporting data. It cited specific macro release estimates (jobless claims: 210K, Philly Fed: -5, GDP prelim: 2.8%, PCE core: 2.3%), provided individual Federal Reserve official quotes by name, tracked international market performance across Asia and Europe, and even noted the potential impact of oil prices on risk sentiment. The level of detail suggested real-time web access actively integrated across the analysis.

The main critique is that Grok’s optimism may have been amplified by the extraordinary quality of the earnings beat. Its 75% probability and $29–$31 closing target imply a net 7–16% gain from the predicted open — a bold call in a week dominated by macro headwinds. The $25.50–$32.50 range is the widest of any chatbot, reflecting awareness of tail risks even within an overall bullish framework.

2. Claude (Anthropic) — The Nuanced Analyst

Opening Prediction: $28.50 | Closing Prediction: $30.00 | Bullish Probability: 58%

Claude predicted the highest opening price of the group at $28.50, reflecting the most aggressive incorporation of after-hours momentum into its gap-up estimate. It identified an implied move of +17% from the $24.19 close, noting that this exceeded the options-implied move of ±12.5% and would therefore require options dealers to delta-hedge by buying additional stock — a mechanical tailwind.

The analysis was distinguished by its structural framing of FIG’s trajectory as a genuine inflection point rather than a “dead-cat bounce”. Revenue growth of 40% with accelerating net retention, guidance 6–8% above consensus, and imminent AI credit monetisation starting March 2026 were cited as “fundamental firepower that can anchor a sustained recovery.”

Claude also provided the most detailed treatment of downside risks. It specifically highlighted the August 2026 lockup expiration — when VCs holding 50%+ of shares would become eligible to sell — as a structural overhang, even within a near-term bullish call. It also flagged CTO Kris Rasmussen’s insider sale of 205,438 shares at $25.01 on 10 February as a concern.

Its 58% bullish probability is notably more measured than Grok’s 75%, explicitly acknowledging that Friday 20 February is “one of the most event-dense trading sessions of 2026” with the simultaneous OPEX, GDP, and PCE release creating a genuine “triple threat.” The predicted closing price of $30.00 (range: $28.00–$33.00) represents a recovery to approximately 24% above the pre-earnings close but still 9% below the IPO price of $33.

3. ChatGPT (OpenAI) — The Cautious Contrarian

Opening Prediction: $26.40 | Closing Prediction: $25.80 | Bearish Probability: 55%

ChatGPT was the only chatbot to produce a net-negative prediction for the period, forecasting that FIG would close the week at $25.80 — below its predicted opening price of $26.40 — giving a 55% probability of a decline. This made it the outlier on the bearish end of the distribution.

The rationale centred on a classic “pop and fade” pattern often seen after earnings gap-ups in volatile, loss-making growth stocks. ChatGPT argued that the elevated opening price would itself be the ceiling, with profit-taking from short-term traders, OPEX-driven pinning effects on Friday, and continued macro headwinds combining to drag the stock lower as the week progressed. Its day-by-day path was unusually granular: it forecast a peak on Day 1 (close: $27.20) followed by stepwise declines through to Wednesday’s close of $25.80.

ChatGPT was also the most explicit about uncertainty and data limitations. Unlike Grok and Claude — which presented comprehensive, source-cited analyses suggesting extensive real-time web access — ChatGPT acknowledged directly that it did not have access to specific options open-interest maps, real-time VIX data, or the current week’s economic calendar. This epistemic transparency, while limiting the richness of the analysis, arguably represents a more honest characterisation of what an AI can and cannot reliably know.

Its predicted trading range of $23.00–$29.80 is notably tighter than Grok’s and centres lower, with the lower bound assuming a gap-fill scenario toward pre-earnings levels. The base case, however, does not predict a full gap-fill: at $25.80 on Wednesday, FIG would still be approximately 6.6% above its 18 February close of $24.19.

4. Perplexity AI — The Transparent Sceptic

Opening Prediction: $24.20 | Closing Prediction: $25.30 | Bullish Probability: 58%

Perplexity’s prediction stands apart from the others for one crucial reason: its opening price estimate of $24.20 was anchored to the prior regular-session close rather than incorporating the after-hours surge. This appears to have been a data-access constraint — the analysis was conducted based on available quote data before fully incorporating the post-earnings jump — which resulted in a materially lower starting point and consequently a more modest predicted range.

Where Perplexity excels is in clearly delineating what it does and does not know. The analysis openly acknowledges that it does not have access to current-week VIX levels, detailed market breadth data, real-time put/call ratios, options open-interest maps, short interest files, or institutional flow statistics. Rather than speculating, it treats these as hard constraints and labels its forecasts explicitly as “scenarios, not data.” This level of epistemic honesty is rare in AI-generated financial analysis.

Despite these data limitations, the analytical framework itself is sound. Perplexity correctly identified the classic “oversold bounce” dynamics, the significance of the volume spike (17.3M shares vs 8.2M average), the importance of the 22–23 support zone and 30 psychological resistance level, and the macro sensitivity of a high-beta growth name to broader NASDAQ sentiment. Its 58% bullish probability aligns with both Claude and Copilot.

The predicted close of $25.30 on a $24.20 open implies a modest +4.5% gain — the smallest net move of any chatbot — which is consistent with a conservative, data-constrained posture. The trading range of $22–$28 is the tightest and lowest of the group.

5. Microsoft Copilot — The Structured Pragmatist

Opening Prediction: $27.50 | Closing Prediction: $28.00 | Bullish Probability: 60%

Copilot occupied the middle ground of the five predictions: a 60% bullish probability, a predicted opening price of $27.50, and a closing price of $28.00 — a modest $0.50 net gain over the week. The analysis was structured, well-organised, and weighted its factors explicitly: earnings and guidance received “high weight,” price action and options flow “high weight,” macro sentiment “medium weight,” and technicals “lower weight” given the short five-day horizon.

Copilot’s predicted opening of $27.50 closely matches the after-hours print of approximately $27–28, suggesting it effectively incorporated the earnings-driven pre-market signals. Its central close estimate of $28.00 (range: $25–$31.50) implies that much of the initial earnings momentum would be sustained but not meaningfully extended — a view of consolidation rather than continuation.

The analysis was notable for its practical, actionable monitoring checklist — a feature the other chatbots largely omitted. It advised readers to watch for whether FIG opened more than 5% below its after-hours level (a signal of immediate profit-taking), to compare volume on up-days versus down-days as a momentum validator, to monitor options sweeps for directional signals, and to track VIX and S&P futures as macro sentiment proxies. This pragmatic, investor-oriented framing suited the TheDayAfterAI News audience well.

Copilot also provided the clearest explicit acknowledgement of the risks that could flip the outcome: a broader market risk-off event, gap-fill profit-taking, negative analyst notes from the earnings call transcript, and unusual put-heavy options sweeps. Its treatment was balanced, if less deep than Grok or Claude on the technical and macro layers.

Comparative Analysis: Where the Chatbots Agreed and Diverged

Points of Consensus

Despite the wide divergence in price targets, all five chatbots agreed on several core themes:

• Strong Earnings as the Primary Driver: Every analysis cited the Q4 2025 beat as the dominant near-term catalyst, with 40% revenue growth and 6–8% guidance upside universally acknowledged.
• Extreme Oversold Technical Conditions: All five noted that FIG’s 83% decline from its IPO-day high had created historically oversold RSI readings, providing a technical foundation for a rebound even within a structural downtrend.
• Friday 20 February as the Key Pivot: OPEX, GDP, and PCE data coinciding on the same session was universally flagged as the highest-risk day of the period.
• Short Interest Dynamics: The 19.4 million shares short (up 31%) and low 1.9 days-to-cover were cited by all five as a modest short-covering tailwind, though not sufficient for a full “squeeze.”
• Macro Headwinds Limiting Upside: Despite varying levels of bullishness, every chatbot acknowledged that the hawkish Fed posture, ongoing SaaSpocalypse sentiment, and volatile broader market environment capped near-term upside potential.

Points of Divergence

Opening Price: The predicted opening prices ranged from $24.20 (Perplexity, appearing to use pre-earnings data) to $28.50 (Claude). Excluding Perplexity’s apparent data constraint, the range narrows to $26.40–$28.50 — still a meaningful $2.10 spread.

Closing Price: The five-day close predictions ranged from $25.30 (Perplexity) to $31.00 (Grok’s upper bound), with central estimates spanning $25.30–$30.00. ChatGPT alone predicted a net loss from open to close.

Bullish Probability: The spread from 45% (ChatGPT’s implied upside probability) to 75% (Grok) is substantial. Three chatbots converged near 58–60%, while Grok and ChatGPT sat as outliers at each extreme.

Analytical Depth: Grok and Claude produced the most extensive analyses — running to thousands of words with specific macro data, RSI/MACD figures, options flow details, and named Fed officials. ChatGPT and Perplexity were more transparent about data limitations. Copilot sat in the middle, structured but less granular on macro and technical layers.

Methodology Comparison

What This Experiment Tells Us About AI and Financial Analysis

This exercise was never intended to identify a “winner.” The actual FIG stock price over the period will tell us which chatbot happened to be closest — but a single outcome cannot validate or invalidate an analytical framework. What this comparison reveals is far more interesting:

1. Different Architectures Produce Different Risk Postures

Even when presented with identical base data, Grok (75% bullish) and ChatGPT (55% bearish) occupied nearly opposite ends of the probability spectrum. This suggests that underlying model training, data access, and risk-calibration assumptions play a larger role in financial prediction outcomes than the raw data inputs themselves.

2. Transparency About Limitations Is a Virtue

Perplexity and ChatGPT were explicit about what they could and could not access. This epistemic honesty is genuinely valuable — an investor who understands the constraints of an AI forecast can weight it appropriately. An analysis that projects false confidence is potentially more dangerous than a transparent acknowledgement of uncertainty.

3. Real-Time Data Access Fundamentally Changes the Output

The most detailed and quantitatively specific analyses — Grok and Claude — appeared to have robust real-time web access, citing specific RSI levels, dark pool data, options open-interest figures, and FOMC official speeches by name. Perplexity’s more conservative output appeared to reflect real-time data constraints. As AI tools continue to evolve, the quality of market data access will likely become the key differentiator in financial analysis quality.

4. Consensus Clustering Is Meaningful

Three chatbots (Claude, Perplexity, Copilot) converged independently on a 58–60% bullish probability. This clustering may be more informative than any individual prediction. When multiple independent AI systems with different architectures and data pipelines arrive at similar probabilistic assessments, the signal-to-noise ratio arguably improves.

5. The Limitations of All AI in Short-Term Market Prediction

It is worth stating plainly: no AI chatbot — regardless of sophistication — can reliably predict short-term stock prices. Markets are shaped by thousands of interacting variables, many of which are unknowable in advance (a surprise PCE figure, an unexpected geopolitical development, a single large institutional trade). These five analyses are best understood as structured frameworks for thinking about risk and probability, not oracles.

Methodology

Each AI chatbot was given an identical prompt requesting a five-day stock price forecast. The models used their own web-search and data-retrieval capabilities; no proprietary data was provided. Responses were collected without modification. Variations in depth, format, and analytical approach reflect each platform’s native capabilities.